an open-source alternative to the dots, bots and muses of the world, run an office of claude code/codex like agents on your laptop, sandboxes or anywhere, uses your existing subscriptions
HTML agent-harnessagent-orchestrationagentsai-agentsautonomous-agentsclaude-codecodexcoding-agentsdesktop-appdeveloper-toolselectrongemini-cliharnessharness-engineeringlocal-firstmemorymulti-agentopencodeorchestrationtypescript
blog: compressing-agent-memory (77 → 78)
Chaitanya Giri9ff332e13f7f
diff --git a/blog/src/posts/compressing-agent-memory.md b/blog/src/posts/compressing-agent-memory.mdnew file mode 100644index 00000000..936af44f--- /dev/null+++ b/blog/src/posts/compressing-agent-memory.md@@ -0,0 +1,127 @@+---+title: "Compressing Agent Memory Without Losing the Original"+description: "How and why to compress an agent's long-term memory: the toolbox, the lossy trap, and keeping a lossless original beside a compact copy."+date: 2026-06-05+category: internals+categoryLabel: Internals+type: Technical+primaryKeyword: "compressing agent memory"+secondaryKeywords: ["agent memory compression", "long-term memory ai agents", "context compression"]+tags: ["Internals", "Memory", "MemPalace", "Multi-Agent"]+author:+ name: Chaitanya Giri+ initials: CG+faq:+ - q: "Why compress an agent's memory instead of using a bigger context window?"+ a: "Because a bigger window carries more noise, costs more per call, and still fills up. The 2026 consensus is to compress what the agent carries forward — fewer, denser tokens of signal — rather than chase a larger window. Compression is what lets a memory store grow for months while the wake-up context an agent loads stays small."+ - q: "Doesn't compressing memory lose information?"+ a: "It can — compression is non-deterministically lossy, and a summary can silently drop the one exception or classification tag that mattered. The fix is to never make the compressed copy the only copy: keep the lossless original on disk and treat the compressed form as a fast-recall accelerator built from it, not a replacement."+ - q: "What actually gets compressed — the stored memory or the prompt?"+ a: "Both are valid, at different layers. You can compact the live conversation as it approaches the window limit, and you can compress long-term stored memory so a session can absorb months of history in a few thousand tokens at wake-up. This post is mostly about the second: durable memory you recall across sessions."+---++<div class="callout tldr"><span class="ic">TL;DR</span><p>An agent's memory grows forever; its context+window does not. The 2026 answer isn't a bigger window — it's <strong>compression</strong>: store dense,+high-signal memory so a session can load months of history in a few thousand tokens. The catch is that+compression is <strong>lossy in unpredictable ways</strong>. The pattern that makes it safe: keep the+<strong>lossless original</strong> on disk and compress a <em>copy</em> for fast recall — never let the+summary become the only record.</p></div>++Every useful agent accumulates memory — decisions, facts, what worked and what didn't. That store only+grows, while the context window an agent can actually read stays fixed. So "memory" for agents is really a+compression problem: how do you let the knowledge base grow for months without dragging all of it back+into a finite, expensive context every session? Here's how to think about compressing agent memory, the+trap to avoid, and the pattern that keeps it safe.++## Why compress at all++The reflex is to reach for a larger context window. In 2026 the better instinct is the opposite. Teams+building the most capable agents have stopped chasing window size and started **deliberately reducing what+the agent carries forward** — because more context means more noise, higher cost per call, and a window+that still fills up. Compression buys three things at once:++- **A bounded wake-up.** An agent should start a session by loading *what it knows* in a small, fixed token+ budget — not by re-reading everything. Compression is what keeps that digest small as the store grows.+- **Lower cost.** Tokens are the bill. Denser memory is a cheaper agent, every call.+- **Better signal.** A tight, distilled memory recalls sharper than a transcript dump. This is the same+ reason [memory hygiene](/blog/keep-agent-semantic-memory-clean/) matters — what you feed the recall layer+ decides what it returns.++## The compression toolbox++A few techniques, cheapest to deepest:++- **Summarization** — roll older history into summaries (hierarchical, so summaries-of-summaries keep the+ oldest material tiny).+- **Selective retention / pruning** — keep mission-critical facts, drop the noise. Not everything is worth+ remembering.+- **Offloading** — push bulky content to disk and keep a pointer plus a short preview, re-reading the full+ thing only on demand.+- **Provider-side compaction** — some APIs compress older conversation turns at a threshold (reductions+ around 80% are reported) so a long session keeps running.+- **Structured distillation** — pack each exchange into a dense, symbolic form. Reported ratios reach ~11x+ on engineering conversations, and purpose-built memory dialects claim far higher.++These differ in where they run, but they share one goal: fewer tokens, same meaning.++## The trap: lossy in ways you can't predict++Compression is not deterministically lossy — it can drop *exactly the wrong thing*. A revenue definition+is simple until it includes "except for multi-year prepayment contracts"; summarize the exception away and+the answer looks clean and computes the wrong number. A table summary that drops the upstream source kills+auditability. A summary that drops a data-classification tag forgets that something was sensitive. You+can't reliably detect these losses after the fact, because each individual summary reads as plausible.++The governance rule that follows: **every compressed chunk should retain a pointer back to its source** —+version and timestamp included — so the system can always route from the dense form back to the exact+original. Which leads to the pattern worth copying.++## The pattern: keep the original, compress a copy++The clean answer is to refuse the false choice between "lossless but huge" and "compact but lossy" — and+keep both. This is how the hive's [semantic memory](/blog/semantic-memory-for-ai-agents/) layer (MemPalace)+is built, and it's the part most worth stealing:++- Each raw memory lives in a **drawer** that holds the **complete original, verbatim** — never summarized,+ never paraphrased. Search returns the exact words you stored.+- When you turn compression on, a **closet** sits beside that drawer holding a **compact, symbolic summary**+ (MemPalace's AAAK dialect reports roughly 30x reduction) that preserves the semantic and relational gist.+- At **wake-up**, an agent loads the compressed closets — absorbing a large span of history in a few+ thousand tokens — while any deep lookup can still pull the lossless drawer.++The harness side is deliberately thin: every agent writes plain+[markdown memory](/blog/markdown-first-agent-memory/), and a background miner feeds those notes into the+shared palace; recall is a `search` or a `wake-up` away. The markdown is the source of truth; the+compressed index is an accelerator built on top of it — never instead of it. That ordering is the whole+trick: **compression buys speed; the preserved original buys correctness.** You get the small wake-up+digest without ever betting your only copy on a summary.++## When not to compress++Compression is a tradeoff, not a default. Skip it (or always keep the original alongside) when the working+set already fits comfortably — compressing tiny context adds latency and lossy risk for no gain; when the+task is **audit- or compliance-critical** and you need the exact source; when the content is full of+**exceptions and edge-cases** a summary tends to flatten; or when data carries **classification or+provenance** that must travel with it. The cross-cutting rule again: never let the compressed copy become+the *only* copy of something you might need to defend or trace.++## FAQ++**Is this the same as RAG?** Related but not the same. Retrieval fetches relevant chunks; compression+changes how densely those chunks are stored and loaded. They compose — you retrieve, and what you retrieve+can be compressed for a smaller, sharper context.++**How much can I compress safely?** As much as you like *if* the lossless original is preserved and+pointer-linked. The compressed form's job is fast recall; correctness lives in the untouched source you can+always fall back to.++**Does compression help cost or just context size?** Both. Fewer tokens loaded per session is directly+fewer tokens billed — compression is one of the cheapest levers on an agent's running cost.++---++Munder Difflin gives every Claude Code agent markdown memory plus+[a shared semantic palace that keeps the lossless original and a compact, recall-fast copy](https://munderdiffl.in/#how)+— so months of context load in a few thousand tokens, locally.+[Download Munder Difflin](https://munderdiffl.in/#install) to give your agents memory that compresses+without forgetting; it's free and open source.diff --git a/docs/blog/append-only-event-log-agents/index.html b/docs/blog/append-only-event-log-agents/index.htmlindex c859da6d..843df6f2 100644--- a/docs/blog/append-only-event-log-agents/index.html+++ b/docs/blog/append-only-event-log-agents/index.html@@ -286,43 +286,43 @@ it’s free and open source.</p><h2>Related in Internals</h2><div class="post-grid"><article class="card">- <a class="thumb t-internals" href="/blog/architecture-two-planes-one-renderer/" aria-hidden="true" tabindex="-1">+ <a class="thumb t-internals" href="/blog/compressing-agent-memory/" aria-hidden="true" tabindex="-1"><span>INTERNALS</span></a><div class="body"><div class="meta">- <time datetime="2026-06-04">Jun 4, 2026</time>+ <time datetime="2026-06-05">Jun 5, 2026</time><span class="dot" aria-hidden="true"></span><span>5 min</span></div>- <h3><a href="/blog/architecture-two-planes-one-renderer/">Munder Difflin's Architecture: Two Data Planes, One Renderer</a></h3>- <p class="dek">A walkthrough of the multi-agent harness architecture: a node-pty terminal plane and a hooks/hive event plane feeding one React + Pixi.js renderer.</p>+ <h3><a href="/blog/compressing-agent-memory/">Compressing Agent Memory Without Losing the Original</a></h3>+ <p class="dek">How and why to compress an agent's long-term memory: the toolbox, the lossy trap, and keeping a lossless original beside a compact copy.</p><div class="foot"><span class="tag kind">Internals</span>- <a class="more" href="/blog/architecture-two-planes-one-renderer/" aria-label="Read: Munder Difflin's Architecture: Two Data Planes, One Renderer">Read →</a>+ <a class="more" href="/blog/compressing-agent-memory/" aria-label="Read: Compressing Agent Memory Without Losing the Original">Read →</a></div></div></article><article class="card">- <a class="thumb t-internals" href="/blog/file-based-coordination-vs-message-queues/" aria-hidden="true" tabindex="-1">+ <a class="thumb t-internals" href="/blog/architecture-two-planes-one-renderer/" aria-hidden="true" tabindex="-1"><span>INTERNALS</span></a><div class="body"><div class="meta"><time datetime="2026-06-04">Jun 4, 2026</time><span class="dot" aria-hidden="true"></span>- <span>4 min</span>+ <span>5 min</span></div>- <h3><a href="/blog/file-based-coordination-vs-message-queues/">File-Based Coordination vs Message Queues for AI Agents</a></h3>- <p class="dek">Why a local agent hive coordinates through plain files, not Redis or RabbitMQ: the zero-ops wins, the real tradeoffs, and where files hit a ceiling.</p>+ <h3><a href="/blog/architecture-two-planes-one-renderer/">Munder Difflin's Architecture: Two Data Planes, One Renderer</a></h3>+ <p class="dek">A walkthrough of the multi-agent harness architecture: a node-pty terminal plane and a hooks/hive event plane feeding one React + Pixi.js renderer.</p><div class="foot"><span class="tag kind">Internals</span>- <a class="more" href="/blog/file-based-coordination-vs-message-queues/" aria-label="Read: File-Based Coordination vs Message Queues for AI Agents">Read →</a>+ <a class="more" href="/blog/architecture-two-planes-one-renderer/" aria-label="Read: Munder Difflin's Architecture: Two Data Planes, One Renderer">Read →</a></div></div></article><article class="card">- <a class="thumb t-internals" href="/blog/observability-for-agent-fleets/" aria-hidden="true" tabindex="-1">+ <a class="thumb t-internals" href="/blog/file-based-coordination-vs-message-queues/" aria-hidden="true" tabindex="-1"><span>INTERNALS</span></a><div class="body">@@ -331,11 +331,11 @@ it’s free and open source.</p><span class="dot" aria-hidden="true"></span><span>4 min</span></div>- <h3><a href="/blog/observability-for-agent-fleets/">Observability for Agent Fleets: Seeing What Your Agents Do</a></h3>- <p class="dek">Agent fleet observability: the four questions your dashboard must answer about who's working, what they're doing, what it costs, and what they know.</p>+ <h3><a href="/blog/file-based-coordination-vs-message-queues/">File-Based Coordination vs Message Queues for AI Agents</a></h3>+ <p class="dek">Why a local agent hive coordinates through plain files, not Redis or RabbitMQ: the zero-ops wins, the real tradeoffs, and where files hit a ceiling.</p><div class="foot"><span class="tag kind">Internals</span>- <a class="more" href="/blog/observability-for-agent-fleets/" aria-label="Read: Observability for Agent Fleets: Seeing What Your Agents Do">Read →</a>+ <a class="more" href="/blog/file-based-coordination-vs-message-queues/" aria-label="Read: File-Based Coordination vs Message Queues for AI Agents">Read →</a></div></div></article>diff --git a/docs/blog/architecture-two-planes-one-renderer/index.html b/docs/blog/architecture-two-planes-one-renderer/index.htmlindex 3fc31948..32c08675 100644--- a/docs/blog/architecture-two-planes-one-renderer/index.html+++ b/docs/blog/architecture-two-planes-one-renderer/index.html@@ -289,25 +289,25 @@ free and open source.</p><h2>Related in Internals</h2><div class="post-grid"><article class="card">- <a class="thumb t-internals" href="/blog/file-based-coordination-vs-message-queues/" aria-hidden="true" tabindex="-1">+ <a class="thumb t-internals" href="/blog/compressing-agent-memory/" aria-hidden="true" tabindex="-1"><span>INTERNALS</span></a><div class="body"><div class="meta">- <time datetime="2026-06-04">Jun 4, 2026</time>+ <time datetime="2026-06-05">Jun 5, 2026</time><span class="dot" aria-hidden="true"></span>- <span>4 min</span>+ <span>5 min</span></div>- <h3><a href="/blog/file-based-coordination-vs-message-queues/">File-Based Coordination vs Message Queues for AI Agents</a></h3>- <p class="dek">Why a local agent hive coordinates through plain files, not Redis or RabbitMQ: the zero-ops wins, the real tradeoffs, and where files hit a ceiling.</p>+ <h3><a href="/blog/compressing-agent-memory/">Compressing Agent Memory Without Losing the Original</a></h3>+ <p class="dek">How and why to compress an agent's long-term memory: the toolbox, the lossy trap, and keeping a lossless original beside a compact copy.</p><div class="foot"><span class="tag kind">Internals</span>- <a class="more" href="/blog/file-based-coordination-vs-message-queues/" aria-label="Read: File-Based Coordination vs Message Queues for AI Agents">Read →</a>+ <a class="more" href="/blog/compressing-agent-memory/" aria-label="Read: Compressing Agent Memory Without Losing the Original">Read →</a></div></div></article><article class="card">- <a class="thumb t-internals" href="/blog/observability-for-agent-fleets/" aria-hidden="true" tabindex="-1">+ <a class="thumb t-internals" href="/blog/file-based-coordination-vs-message-queues/" aria-hidden="true" tabindex="-1"><span>INTERNALS</span></a><div class="body">@@ -316,29 +316,29 @@ free and open source.</p><span class="dot" aria-hidden="true"></span><span>4 min</span></div>- <h3><a href="/blog/observability-for-agent-fleets/">Observability for Agent Fleets: Seeing What Your Agents Do</a></h3>- <p class="dek">Agent fleet observability: the four questions your dashboard must answer about who's working, what they're doing, what it costs, and what they know.</p>+ <h3><a href="/blog/file-based-coordination-vs-message-queues/">File-Based Coordination vs Message Queues for AI Agents</a></h3>+ <p class="dek">Why a local agent hive coordinates through plain files, not Redis or RabbitMQ: the zero-ops wins, the real tradeoffs, and where files hit a ceiling.</p><div class="foot"><span class="tag kind">Internals</span>- <a class="more" href="/blog/observability-for-agent-fleets/" aria-label="Read: Observability for Agent Fleets: Seeing What Your Agents Do">Read →</a>+ <a class="more" href="/blog/file-based-coordination-vs-message-queues/" aria-label="Read: File-Based Coordination vs Message Queues for AI Agents">Read →</a></div></div></article><article class="card">- <a class="thumb t-internals" href="/blog/recovering-from-agent-failures/" aria-hidden="true" tabindex="-1">+ <a class="thumb t-internals" href="/blog/observability-for-agent-fleets/" aria-hidden="true" tabindex="-1"><span>INTERNALS</span></a><div class="body"><div class="meta"><time datetime="2026-06-04">Jun 4, 2026</time><span class="dot" aria-hidden="true"></span>- <span>6 min</span>+ <span>4 min</span></div>- <h3><a href="/blog/recovering-from-agent-failures/">Recovering From Agent Failures: Resilience Patterns for a Hive</a></h3>- <p class="dek">How multi-agent systems survive partial failure: retry with stale-lock recovery, poison-message quarantine, hop-cap circuit breaking, idempotent handling.</p>+ <h3><a href="/blog/observability-for-agent-fleets/">Observability for Agent Fleets: Seeing What Your Agents Do</a></h3>+ <p class="dek">Agent fleet observability: the four questions your dashboard must answer about who's working, what they're doing, what it costs, and what they know.</p><div class="foot"><span class="tag kind">Internals</span>- <a class="more" href="/blog/recovering-from-agent-failures/" aria-label="Read: Recovering From Agent Failures: Resilience Patterns for a Hive">Read →</a>+ <a class="more" href="/blog/observability-for-agent-fleets/" aria-label="Read: Observability for Agent Fleets: Seeing What Your Agents Do">Read →</a></div></div></article>diff --git a/docs/blog/atomic-file-mailboxes-for-agents/index.html b/docs/blog/atomic-file-mailboxes-for-agents/index.htmlindex d132d527..ce59d1dc 100644--- a/docs/blog/atomic-file-mailboxes-for-agents/index.html+++ b/docs/blog/atomic-file-mailboxes-for-agents/index.html@@ -301,43 +301,43 @@ the office floor; it’s free and open source.</p><h2>Related in Internals</h2><div class="post-grid"><article class="card">- <a class="thumb t-internals" href="/blog/architecture-two-planes-one-renderer/" aria-hidden="true" tabindex="-1">+ <a class="thumb t-internals" href="/blog/compressing-agent-memory/" aria-hidden="true" tabindex="-1"><span>INTERNALS</span></a><div class="body"><div class="meta">- <time datetime="2026-06-04">Jun 4, 2026</time>+ <time datetime="2026-06-05">Jun 5, 2026</time><span class="dot" aria-hidden="true"></span><span>5 min</span></div>- <h3><a href="/blog/architecture-two-planes-one-renderer/">Munder Difflin's Architecture: Two Data Planes, One Renderer</a></h3>- <p class="dek">A walkthrough of the multi-agent harness architecture: a node-pty terminal plane and a hooks/hive event plane feeding one React + Pixi.js renderer.</p>+ <h3><a href="/blog/compressing-agent-memory/">Compressing Agent Memory Without Losing the Original</a></h3>+ <p class="dek">How and why to compress an agent's long-term memory: the toolbox, the lossy trap, and keeping a lossless original beside a compact copy.</p><div class="foot"><span class="tag kind">Internals</span>- <a class="more" href="/blog/architecture-two-planes-one-renderer/" aria-label="Read: Munder Difflin's Architecture: Two Data Planes, One Renderer">Read →</a>+ <a class="more" href="/blog/compressing-agent-memory/" aria-label="Read: Compressing Agent Memory Without Losing the Original">Read →</a></div></div></article><article class="card">- <a class="thumb t-internals" href="/blog/file-based-coordination-vs-message-queues/" aria-hidden="true" tabindex="-1">+ <a class="thumb t-internals" href="/blog/architecture-two-planes-one-renderer/" aria-hidden="true" tabindex="-1"><span>INTERNALS</span></a><div class="body"><div class="meta"><time datetime="2026-06-04">Jun 4, 2026</time><span class="dot" aria-hidden="true"></span>- <span>4 min</span>+ <span>5 min</span></div>- <h3><a href="/blog/file-based-coordination-vs-message-queues/">File-Based Coordination vs Message Queues for AI Agents</a></h3>- <p class="dek">Why a local agent hive coordinates through plain files, not Redis or RabbitMQ: the zero-ops wins, the real tradeoffs, and where files hit a ceiling.</p>+ <h3><a href="/blog/architecture-two-planes-one-renderer/">Munder Difflin's Architecture: Two Data Planes, One Renderer</a></h3>+ <p class="dek">A walkthrough of the multi-agent harness architecture: a node-pty terminal plane and a hooks/hive event plane feeding one React + Pixi.js renderer.</p><div class="foot"><span class="tag kind">Internals</span>- <a class="more" href="/blog/file-based-coordination-vs-message-queues/" aria-label="Read: File-Based Coordination vs Message Queues for AI Agents">Read →</a>+ <a class="more" href="/blog/architecture-two-planes-one-renderer/" aria-label="Read: Munder Difflin's Architecture: Two Data Planes, One Renderer">Read →</a></div></div></article><article class="card">- <a class="thumb t-internals" href="/blog/observability-for-agent-fleets/" aria-hidden="true" tabindex="-1">+ <a class="thumb t-internals" href="/blog/file-based-coordination-vs-message-queues/" aria-hidden="true" tabindex="-1"><span>INTERNALS</span></a><div class="body">@@ -346,11 +346,11 @@ the office floor; it’s free and open source.</p><span class="dot" aria-hidden="true"></span><span>4 min</span></div>- <h3><a href="/blog/observability-for-agent-fleets/">Observability for Agent Fleets: Seeing What Your Agents Do</a></h3>- <p class="dek">Agent fleet observability: the four questions your dashboard must answer about who's working, what they're doing, what it costs, and what they know.</p>+ <h3><a href="/blog/file-based-coordination-vs-message-queues/">File-Based Coordination vs Message Queues for AI Agents</a></h3>+ <p class="dek">Why a local agent hive coordinates through plain files, not Redis or RabbitMQ: the zero-ops wins, the real tradeoffs, and where files hit a ceiling.</p><div class="foot"><span class="tag kind">Internals</span>- <a class="more" href="/blog/observability-for-agent-fleets/" aria-label="Read: Observability for Agent Fleets: Seeing What Your Agents Do">Read →</a>+ <a class="more" href="/blog/file-based-coordination-vs-message-queues/" aria-label="Read: File-Based Coordination vs Message Queues for AI Agents">Read →</a></div></div></article>diff --git a/docs/blog/building-a-terminal-ui-xterm-node-pty/index.html b/docs/blog/building-a-terminal-ui-xterm-node-pty/index.htmlindex f2cec0c4..1e271195 100644--- a/docs/blog/building-a-terminal-ui-xterm-node-pty/index.html+++ b/docs/blog/building-a-terminal-ui-xterm-node-pty/index.html@@ -344,43 +344,43 @@ to watch many real terminals at once; it’s free and open source.</p><h2>Related in Internals</h2><div class="post-grid"><article class="card">- <a class="thumb t-internals" href="/blog/architecture-two-planes-one-renderer/" aria-hidden="true" tabindex="-1">+ <a class="thumb t-internals" href="/blog/compressing-agent-memory/" aria-hidden="true" tabindex="-1"><span>INTERNALS</span></a><div class="body"><div class="meta">- <time datetime="2026-06-04">Jun 4, 2026</time>+ <time datetime="2026-06-05">Jun 5, 2026</time><span class="dot" aria-hidden="true"></span><span>5 min</span></div>- <h3><a href="/blog/architecture-two-planes-one-renderer/">Munder Difflin's Architecture: Two Data Planes, One Renderer</a></h3>- <p class="dek">A walkthrough of the multi-agent harness architecture: a node-pty terminal plane and a hooks/hive event plane feeding one React + Pixi.js renderer.</p>+ <h3><a href="/blog/compressing-agent-memory/">Compressing Agent Memory Without Losing the Original</a></h3>+ <p class="dek">How and why to compress an agent's long-term memory: the toolbox, the lossy trap, and keeping a lossless original beside a compact copy.</p><div class="foot"><span class="tag kind">Internals</span>- <a class="more" href="/blog/architecture-two-planes-one-renderer/" aria-label="Read: Munder Difflin's Architecture: Two Data Planes, One Renderer">Read →</a>+ <a class="more" href="/blog/compressing-agent-memory/" aria-label="Read: Compressing Agent Memory Without Losing the Original">Read →</a></div></div></article><article class="card">- <a class="thumb t-internals" href="/blog/file-based-coordination-vs-message-queues/" aria-hidden="true" tabindex="-1">+ <a class="thumb t-internals" href="/blog/architecture-two-planes-one-renderer/" aria-hidden="true" tabindex="-1"><span>INTERNALS</span></a><div class="body"><div class="meta"><time datetime="2026-06-04">Jun 4, 2026</time><span class="dot" aria-hidden="true"></span>- <span>4 min</span>+ <span>5 min</span></div>- <h3><a href="/blog/file-based-coordination-vs-message-queues/">File-Based Coordination vs Message Queues for AI Agents</a></h3>- <p class="dek">Why a local agent hive coordinates through plain files, not Redis or RabbitMQ: the zero-ops wins, the real tradeoffs, and where files hit a ceiling.</p>+ <h3><a href="/blog/architecture-two-planes-one-renderer/">Munder Difflin's Architecture: Two Data Planes, One Renderer</a></h3>+ <p class="dek">A walkthrough of the multi-agent harness architecture: a node-pty terminal plane and a hooks/hive event plane feeding one React + Pixi.js renderer.</p><div class="foot"><span class="tag kind">Internals</span>- <a class="more" href="/blog/file-based-coordination-vs-message-queues/" aria-label="Read: File-Based Coordination vs Message Queues for AI Agents">Read →</a>+ <a class="more" href="/blog/architecture-two-planes-one-renderer/" aria-label="Read: Munder Difflin's Architecture: Two Data Planes, One Renderer">Read →</a></div></div></article><article class="card">- <a class="thumb t-internals" href="/blog/observability-for-agent-fleets/" aria-hidden="true" tabindex="-1">+ <a class="thumb t-internals" href="/blog/file-based-coordination-vs-message-queues/" aria-hidden="true" tabindex="-1"><span>INTERNALS</span></a><div class="body">@@ -389,11 +389,11 @@ to watch many real terminals at once; it’s free and open source.</p><span class="dot" aria-hidden="true"></span><span>4 min</span></div>- <h3><a href="/blog/observability-for-agent-fleets/">Observability for Agent Fleets: Seeing What Your Agents Do</a></h3>- <p class="dek">Agent fleet observability: the four questions your dashboard must answer about who's working, what they're doing, what it costs, and what they know.</p>+ <h3><a href="/blog/file-based-coordination-vs-message-queues/">File-Based Coordination vs Message Queues for AI Agents</a></h3>+ <p class="dek">Why a local agent hive coordinates through plain files, not Redis or RabbitMQ: the zero-ops wins, the real tradeoffs, and where files hit a ceiling.</p><div class="foot"><span class="tag kind">Internals</span>- <a class="more" href="/blog/observability-for-agent-fleets/" aria-label="Read: Observability for Agent Fleets: Seeing What Your Agents Do">Read →</a>+ <a class="more" href="/blog/file-based-coordination-vs-message-queues/" aria-label="Read: File-Based Coordination vs Message Queues for AI Agents">Read →</a></div></div></article>diff --git a/docs/blog/building-an-ai-office-floor/index.html b/docs/blog/building-an-ai-office-floor/index.htmlindex 457fc422..2f80ba9c 100644--- a/docs/blog/building-an-ai-office-floor/index.html+++ b/docs/blog/building-an-ai-office-floor/index.html@@ -289,43 +289,43 @@ techniques — the hard part is wiring them to <em>real</em> agent events, not t<h2>Related in Internals</h2><div class="post-grid"><article class="card">- <a class="thumb t-internals" href="/blog/architecture-two-planes-one-renderer/" aria-hidden="true" tabindex="-1">+ <a class="thumb t-internals" href="/blog/compressing-agent-memory/" aria-hidden="true" tabindex="-1"><span>INTERNALS</span></a><div class="body"><div class="meta">- <time datetime="2026-06-04">Jun 4, 2026</time>+ <time datetime="2026-06-05">Jun 5, 2026</time><span class="dot" aria-hidden="true"></span><span>5 min</span></div>- <h3><a href="/blog/architecture-two-planes-one-renderer/">Munder Difflin's Architecture: Two Data Planes, One Renderer</a></h3>- <p class="dek">A walkthrough of the multi-agent harness architecture: a node-pty terminal plane and a hooks/hive event plane feeding one React + Pixi.js renderer.</p>+ <h3><a href="/blog/compressing-agent-memory/">Compressing Agent Memory Without Losing the Original</a></h3>+ <p class="dek">How and why to compress an agent's long-term memory: the toolbox, the lossy trap, and keeping a lossless original beside a compact copy.</p><div class="foot"><span class="tag kind">Internals</span>- <a class="more" href="/blog/architecture-two-planes-one-renderer/" aria-label="Read: Munder Difflin's Architecture: Two Data Planes, One Renderer">Read →</a>+ <a class="more" href="/blog/compressing-agent-memory/" aria-label="Read: Compressing Agent Memory Without Losing the Original">Read →</a></div></div></article><article class="card">- <a class="thumb t-internals" href="/blog/file-based-coordination-vs-message-queues/" aria-hidden="true" tabindex="-1">+ <a class="thumb t-internals" href="/blog/architecture-two-planes-one-renderer/" aria-hidden="true" tabindex="-1"><span>INTERNALS</span></a><div class="body"><div class="meta"><time datetime="2026-06-04">Jun 4, 2026</time><span class="dot" aria-hidden="true"></span>- <span>4 min</span>+ <span>5 min</span></div>- <h3><a href="/blog/file-based-coordination-vs-message-queues/">File-Based Coordination vs Message Queues for AI Agents</a></h3>- <p class="dek">Why a local agent hive coordinates through plain files, not Redis or RabbitMQ: the zero-ops wins, the real tradeoffs, and where files hit a ceiling.</p>+ <h3><a href="/blog/architecture-two-planes-one-renderer/">Munder Difflin's Architecture: Two Data Planes, One Renderer</a></h3>+ <p class="dek">A walkthrough of the multi-agent harness architecture: a node-pty terminal plane and a hooks/hive event plane feeding one React + Pixi.js renderer.</p><div class="foot"><span class="tag kind">Internals</span>- <a class="more" href="/blog/file-based-coordination-vs-message-queues/" aria-label="Read: File-Based Coordination vs Message Queues for AI Agents">Read →</a>+ <a class="more" href="/blog/architecture-two-planes-one-renderer/" aria-label="Read: Munder Difflin's Architecture: Two Data Planes, One Renderer">Read →</a></div></div></article><article class="card">- <a class="thumb t-internals" href="/blog/observability-for-agent-fleets/" aria-hidden="true" tabindex="-1">+ <a class="thumb t-internals" href="/blog/file-based-coordination-vs-message-queues/" aria-hidden="true" tabindex="-1"><span>INTERNALS</span></a><div class="body">@@ -334,11 +334,11 @@ techniques — the hard part is wiring them to <em>real</em> agent events, not t<span class="dot" aria-hidden="true"></span><span>4 min</span></div>- <h3><a href="/blog/observability-for-agent-fleets/">Observability for Agent Fleets: Seeing What Your Agents Do</a></h3>- <p class="dek">Agent fleet observability: the four questions your dashboard must answer about who's working, what they're doing, what it costs, and what they know.</p>+ <h3><a href="/blog/file-based-coordination-vs-message-queues/">File-Based Coordination vs Message Queues for AI Agents</a></h3>+ <p class="dek">Why a local agent hive coordinates through plain files, not Redis or RabbitMQ: the zero-ops wins, the real tradeoffs, and where files hit a ceiling.</p><div class="foot"><span class="tag kind">Internals</span>- <a class="more" href="/blog/observability-for-agent-fleets/" aria-label="Read: Observability for Agent Fleets: Seeing What Your Agents Do">Read →</a>+ <a class="more" href="/blog/file-based-coordination-vs-message-queues/" aria-label="Read: File-Based Coordination vs Message Queues for AI Agents">Read →</a></div></div></article>diff --git a/docs/blog/claude-code-hooks-explained/index.html b/docs/blog/claude-code-hooks-explained/index.htmlindex 6ca63695..7c64fbbc 100644--- a/docs/blog/claude-code-hooks-explained/index.html+++ b/docs/blog/claude-code-hooks-explained/index.html@@ -303,43 +303,43 @@ free and open source.</p><h2>Related in Internals</h2><div class="post-grid"><article class="card">- <a class="thumb t-internals" href="/blog/architecture-two-planes-one-renderer/" aria-hidden="true" tabindex="-1">+ <a class="thumb t-internals" href="/blog/compressing-agent-memory/" aria-hidden="true" tabindex="-1"><span>INTERNALS</span></a><div class="body"><div class="meta">- <time datetime="2026-06-04">Jun 4, 2026</time>+ <time datetime="2026-06-05">Jun 5, 2026</time><span class="dot" aria-hidden="true"></span><span>5 min</span></div>- <h3><a href="/blog/architecture-two-planes-one-renderer/">Munder Difflin's Architecture: Two Data Planes, One Renderer</a></h3>- <p class="dek">A walkthrough of the multi-agent harness architecture: a node-pty terminal plane and a hooks/hive event plane feeding one React + Pixi.js renderer.</p>+ <h3><a href="/blog/compressing-agent-memory/">Compressing Agent Memory Without Losing the Original</a></h3>+ <p class="dek">How and why to compress an agent's long-term memory: the toolbox, the lossy trap, and keeping a lossless original beside a compact copy.</p><div class="foot"><span class="tag kind">Internals</span>- <a class="more" href="/blog/architecture-two-planes-one-renderer/" aria-label="Read: Munder Difflin's Architecture: Two Data Planes, One Renderer">Read →</a>+ <a class="more" href="/blog/compressing-agent-memory/" aria-label="Read: Compressing Agent Memory Without Losing the Original">Read →</a></div></div></article><article class="card">- <a class="thumb t-internals" href="/blog/file-based-coordination-vs-message-queues/" aria-hidden="true" tabindex="-1">+ <a class="thumb t-internals" href="/blog/architecture-two-planes-one-renderer/" aria-hidden="true" tabindex="-1"><span>INTERNALS</span></a><div class="body"><div class="meta"><time datetime="2026-06-04">Jun 4, 2026</time><span class="dot" aria-hidden="true"></span>- <span>4 min</span>+ <span>5 min</span></div>- <h3><a href="/blog/file-based-coordination-vs-message-queues/">File-Based Coordination vs Message Queues for AI Agents</a></h3>- <p class="dek">Why a local agent hive coordinates through plain files, not Redis or RabbitMQ: the zero-ops wins, the real tradeoffs, and where files hit a ceiling.</p>+ <h3><a href="/blog/architecture-two-planes-one-renderer/">Munder Difflin's Architecture: Two Data Planes, One Renderer</a></h3>+ <p class="dek">A walkthrough of the multi-agent harness architecture: a node-pty terminal plane and a hooks/hive event plane feeding one React + Pixi.js renderer.</p><div class="foot"><span class="tag kind">Internals</span>- <a class="more" href="/blog/file-based-coordination-vs-message-queues/" aria-label="Read: File-Based Coordination vs Message Queues for AI Agents">Read →</a>+ <a class="more" href="/blog/architecture-two-planes-one-renderer/" aria-label="Read: Munder Difflin's Architecture: Two Data Planes, One Renderer">Read →</a></div></div></article><article class="card">- <a class="thumb t-internals" href="/blog/observability-for-agent-fleets/" aria-hidden="true" tabindex="-1">+ <a class="thumb t-internals" href="/blog/file-based-coordination-vs-message-queues/" aria-hidden="true" tabindex="-1"><span>INTERNALS</span></a><div class="body">@@ -348,11 +348,11 @@ free and open source.</p><span class="dot" aria-hidden="true"></span><span>4 min</span></div>- <h3><a href="/blog/observability-for-agent-fleets/">Observability for Agent Fleets: Seeing What Your Agents Do</a></h3>- <p class="dek">Agent fleet observability: the four questions your dashboard must answer about who's working, what they're doing, what it costs, and what they know.</p>+ <h3><a href="/blog/file-based-coordination-vs-message-queues/">File-Based Coordination vs Message Queues for AI Agents</a></h3>+ <p class="dek">Why a local agent hive coordinates through plain files, not Redis or RabbitMQ: the zero-ops wins, the real tradeoffs, and where files hit a ceiling.</p><div class="foot"><span class="tag kind">Internals</span>- <a class="more" href="/blog/observability-for-agent-fleets/" aria-label="Read: Observability for Agent Fleets: Seeing What Your Agents Do">Read →</a>+ <a class="more" href="/blog/file-based-coordination-vs-message-queues/" aria-label="Read: File-Based Coordination vs Message Queues for AI Agents">Read →</a></div></div></article>diff --git a/docs/blog/cline-vs-munder-difflin/index.html b/docs/blog/cline-vs-munder-difflin/index.htmlindex f25fbd0e..4d3f0833 100644--- a/docs/blog/cline-vs-munder-difflin/index.html+++ b/docs/blog/cline-vs-munder-difflin/index.html@@ -286,7 +286,7 @@ licensed.</p></div><footer class="post-foot wrap"><div class="prevnext">- <a class="pn prev" href="/blog/generative-engine-optimization-techniques/"><span class="k">← Older</span><span class="t">Generative Engine Optimization: The Techniques the Research Backs</span></a>+ <a class="pn prev" href="/blog/compressing-agent-memory/"><span class="k">← Older</span><span class="t">Compressing Agent Memory Without Losing the Original</span></a><span></span></div><section class="related">diff --git a/docs/blog/compressing-agent-memory/index.html b/docs/blog/compressing-agent-memory/index.htmlnew file mode 100644index 00000000..7276440b--- /dev/null+++ b/docs/blog/compressing-agent-memory/index.html@@ -0,0 +1,370 @@+<!doctype html>+<html lang="en">+<head>+<meta charset="UTF-8" />+<meta name="viewport" content="width=device-width, initial-scale=1.0" />+<title>Compressing Agent Memory Without Losing the Original — Munder Difflin Blog</title>+<meta name="description" content="How and why to compress an agent's long-term memory: the toolbox, the lossy trap, and keeping a lossless original beside a compact copy." />+<link rel="canonical" href="https://munderdiffl.in/blog/compressing-agent-memory/" />++<link rel="icon" type="image/png" href="https://munderdiffl.in/logo.png" />+<link rel="apple-touch-icon" href="https://munderdiffl.in/logo.png" />+<meta name="theme-color" content="#F5F2E8" />++<!-- Open Graph -->+<meta property="og:site_name" content="Munder Difflin Blog" />+<meta property="og:type" content="article" />+<meta property="og:title" content="Compressing Agent Memory Without Losing the Original — Munder Difflin Blog" />+<meta property="og:description" content="How and why to compress an agent's long-term memory: the toolbox, the lossy trap, and keeping a lossless original beside a compact copy." />+<meta property="og:url" content="https://munderdiffl.in/blog/compressing-agent-memory/" />+<meta property="og:image" content="https://munderdiffl.in/media/og.png" />+<meta property="og:locale" content="en_US" />+<meta property="article:published_time" content="2026-06-05T00:00:00.000Z" />+<meta property="article:author" content="Chaitanya Giri" />+<meta property="article:tag" content="Internals" />+<meta property="article:tag" content="Memory" />+<meta property="article:tag" content="MemPalace" />+<meta property="article:tag" content="Multi-Agent" />+++<!-- Twitter -->+<meta name="twitter:card" content="summary_large_image" />+<meta name="twitter:title" content="Compressing Agent Memory Without Losing the Original — Munder Difflin Blog" />+<meta name="twitter:description" content="How and why to compress an agent's long-term memory: the toolbox, the lossy trap, and keeping a lossless original beside a compact copy." />+<meta name="twitter:image" content="https://munderdiffl.in/media/og.png" />++<link rel="alternate" type="application/atom+xml" title="Munder Difflin Blog" href="https://munderdiffl.in/blog/feed.xml" />+<link rel="preconnect" href="https://fonts.googleapis.com" />+<link rel="preconnect" href="https://fonts.gstatic.com" crossorigin />+<link href="https://fonts.googleapis.com/css2?family=Geist:wght@400;500;600&family=JetBrains+Mono:wght@400;500;600;700&display=swap" rel="stylesheet" />+<link rel="stylesheet" href="/blog/assets/blog.css" />++<script type="application/ld+json">+{+ "@context": "https://schema.org",+ "@type": "BlogPosting",+ "headline": "Compressing Agent Memory Without Losing the Original",+ "description": "How and why to compress an agent's long-term memory: the toolbox, the lossy trap, and keeping a lossless original beside a compact copy.",+ "image": ["https://munderdiffl.in/media/og.png"],+ "datePublished": "2026-06-05T00:00:00.000Z",+ "dateModified": "2026-06-05T00:00:00.000Z",+ "author": { "@type": "Person", "name": "Chaitanya Giri" },+ "publisher": {+ "@type": "Organization",+ "name": "Munder Difflin",+ "logo": { "@type": "ImageObject", "url": "https://munderdiffl.in/logo.png" }+ },+ "mainEntityOfPage": { "@type": "WebPage", "@id": "https://munderdiffl.in/blog/compressing-agent-memory/" },+ "keywords": "compressing agent memory, agent memory compression, long-term memory ai agents, context compression",+ "articleSection": "Internals",+ "inLanguage": "en-US",+ "url": "https://munderdiffl.in/blog/compressing-agent-memory/"+}+</script>+<script type="application/ld+json">+{+ "@context": "https://schema.org",+ "@type": "BreadcrumbList",+ "itemListElement": [+ { "@type": "ListItem", "position": 1, "name": "Home", "item": "https://munderdiffl.in/" },+ { "@type": "ListItem", "position": 2, "name": "Blog", "item": "https://munderdiffl.in/blog/" },+ { "@type": "ListItem", "position": 3, "name": "Compressing Agent Memory Without Losing the Original", "item": "https://munderdiffl.in/blog/compressing-agent-memory/" }+ ]+}+</script>++<script type="application/ld+json">+{+ "@context": "https://schema.org",+ "@type": "FAQPage",+ "mainEntity": [+ { "@type": "Question", "name": "Why compress an agent's memory instead of using a bigger context window?", "acceptedAnswer": { "@type": "Answer", "text": "Because a bigger window carries more noise, costs more per call, and still fills up. The 2026 consensus is to compress what the agent carries forward — fewer, denser tokens of signal — rather than chase a larger window. Compression is what lets a memory store grow for months while the wake-up context an agent loads stays small." } },+ { "@type": "Question", "name": "Doesn't compressing memory lose information?", "acceptedAnswer": { "@type": "Answer", "text": "It can — compression is non-deterministically lossy, and a summary can silently drop the one exception or classification tag that mattered. The fix is to never make the compressed copy the only copy: keep the lossless original on disk and treat the compressed form as a fast-recall accelerator built from it, not a replacement." } },+ { "@type": "Question", "name": "What actually gets compressed — the stored memory or the prompt?", "acceptedAnswer": { "@type": "Answer", "text": "Both are valid, at different layers. You can compact the live conversation as it approaches the window limit, and you can compress long-term stored memory so a session can absorb months of history in a few thousand tokens at wake-up. This post is mostly about the second: durable memory you recall across sessions." } }++ ]+}+</script>+++</head>+<body>+<a class="skip-link" href="#main">Skip to content</a>+<nav id="nav">+ <a class="brand" href="/blog/" aria-label="Munder Difflin Blog home">+ <span class="brand-mark">MD</span>+ <span class="brand-word">MUNDER DIFFLIN<span class="sep">/</span><span class="blog">BLOG</span></span>+ </a>+ <span class="spacer"></span>+ <div class="links">+ <a href="/blog/">Latest</a>+ <a href="/blog/topics/">Topics</a>+ <a href="/blog/about/">About</a>+ <a href="https://munderdiffl.in">↗ Product</a>+ </div>+ <div class="nav-actions">+ <a class="btn sm" href="https://munderdiffl.in/blog/feed.xml" aria-label="RSS feed">⤵ RSS</a>+ <a class="btn sm primary" href="https://munderdiffl.in/#install">⤓ Download</a>+ </div>+</nav>++<main id="main">++<article>+ <header class="post-head wrap">+ <nav class="crumbs" aria-label="Breadcrumb">+ <a href="/blog/">Blog</a>+ <span>/</span>+ <a href="/blog/topics/internals/">Internals</a>+ </nav>+ <h1>Compressing Agent Memory Without Losing the Original</h1>+ <p class="dek">How and why to compress an agent's long-term memory: the toolbox, the lossy trap, and keeping a lossless original beside a compact copy.</p>+ <div class="byline">+ <span class="av" aria-hidden="true">CG</span>+ <strong>Chaitanya Giri</strong>+ <span class="dot" aria-hidden="true">·</span>+ <time datetime="2026-06-05">Jun 5, 2026</time>+ <span class="dot" aria-hidden="true">·</span>+ <span>5 min read</span>+ <span class="tag kind">Internals</span>+ </div>+ </header>++ <div class="post-wrap wrap">+ <div class="prose">+ <div class="callout tldr"><span class="ic">TL;DR</span><p>An agent's memory grows forever; its context+window does not. The 2026 answer isn't a bigger window — it's <strong>compression</strong>: store dense,+high-signal memory so a session can load months of history in a few thousand tokens. The catch is that+compression is <strong>lossy in unpredictable ways</strong>. The pattern that makes it safe: keep the+<strong>lossless original</strong> on disk and compress a <em>copy</em> for fast recall — never let the+summary become the only record.</p></div>+<p>Every useful agent accumulates memory — decisions, facts, what worked and what didn’t. That store only+grows, while the context window an agent can actually read stays fixed. So “memory” for agents is really a+compression problem: how do you let the knowledge base grow for months without dragging all of it back+into a finite, expensive context every session? Here’s how to think about compressing agent memory, the+trap to avoid, and the pattern that keeps it safe.</p>+<h2 id="why-compress-at-all" tabindex="-1">Why compress at all <a class="anchor" href="#why-compress-at-all" aria-hidden="true">#</a></h2>+<p>The reflex is to reach for a larger context window. In 2026 the better instinct is the opposite. Teams+building the most capable agents have stopped chasing window size and started <strong>deliberately reducing what+the agent carries forward</strong> — because more context means more noise, higher cost per call, and a window+that still fills up. Compression buys three things at once:</p>+<ul>+<li><strong>A bounded wake-up.</strong> An agent should start a session by loading <em>what it knows</em> in a small, fixed token+budget — not by re-reading everything. Compression is what keeps that digest small as the store grows.</li>+<li><strong>Lower cost.</strong> Tokens are the bill. Denser memory is a cheaper agent, every call.</li>+<li><strong>Better signal.</strong> A tight, distilled memory recalls sharper than a transcript dump. This is the same+reason <a href="/blog/keep-agent-semantic-memory-clean/">memory hygiene</a> matters — what you feed the recall layer+decides what it returns.</li>+</ul>+<h2 id="the-compression-toolbox" tabindex="-1">The compression toolbox <a class="anchor" href="#the-compression-toolbox" aria-hidden="true">#</a></h2>+<p>A few techniques, cheapest to deepest:</p>+<ul>+<li><strong>Summarization</strong> — roll older history into summaries (hierarchical, so summaries-of-summaries keep the+oldest material tiny).</li>+<li><strong>Selective retention / pruning</strong> — keep mission-critical facts, drop the noise. Not everything is worth+remembering.</li>+<li><strong>Offloading</strong> — push bulky content to disk and keep a pointer plus a short preview, re-reading the full+thing only on demand.</li>+<li><strong>Provider-side compaction</strong> — some APIs compress older conversation turns at a threshold (reductions+around 80% are reported) so a long session keeps running.</li>+<li><strong>Structured distillation</strong> — pack each exchange into a dense, symbolic form. Reported ratios reach ~11x+on engineering conversations, and purpose-built memory dialects claim far higher.</li>+</ul>+<p>These differ in where they run, but they share one goal: fewer tokens, same meaning.</p>+<h2 id="the-trap-lossy-in-ways-you-cant-predict" tabindex="-1">The trap: lossy in ways you can’t predict <a class="anchor" href="#the-trap-lossy-in-ways-you-cant-predict" aria-hidden="true">#</a></h2>+<p>Compression is not deterministically lossy — it can drop <em>exactly the wrong thing</em>. A revenue definition+is simple until it includes “except for multi-year prepayment contracts”; summarize the exception away and+the answer looks clean and computes the wrong number. A table summary that drops the upstream source kills+auditability. A summary that drops a data-classification tag forgets that something was sensitive. You+can’t reliably detect these losses after the fact, because each individual summary reads as plausible.</p>+<p>The governance rule that follows: <strong>every compressed chunk should retain a pointer back to its source</strong> —+version and timestamp included — so the system can always route from the dense form back to the exact+original. Which leads to the pattern worth copying.</p>+<h2 id="the-pattern-keep-the-original-compress-a-copy" tabindex="-1">The pattern: keep the original, compress a copy <a class="anchor" href="#the-pattern-keep-the-original-compress-a-copy" aria-hidden="true">#</a></h2>+<p>The clean answer is to refuse the false choice between “lossless but huge” and “compact but lossy” — and+keep both. This is how the hive’s <a href="/blog/semantic-memory-for-ai-agents/">semantic memory</a> layer (MemPalace)+is built, and it’s the part most worth stealing:</p>+<ul>+<li>Each raw memory lives in a <strong>drawer</strong> that holds the <strong>complete original, verbatim</strong> — never summarized,+never paraphrased. Search returns the exact words you stored.</li>+<li>When you turn compression on, a <strong>closet</strong> sits beside that drawer holding a <strong>compact, symbolic summary</strong>+(MemPalace’s AAAK dialect reports roughly 30x reduction) that preserves the semantic and relational gist.</li>+<li>At <strong>wake-up</strong>, an agent loads the compressed closets — absorbing a large span of history in a few+thousand tokens — while any deep lookup can still pull the lossless drawer.</li>+</ul>+<p>The harness side is deliberately thin: every agent writes plain+<a href="/blog/markdown-first-agent-memory/">markdown memory</a>, and a background miner feeds those notes into the+shared palace; recall is a <code>search</code> or a <code>wake-up</code> away. The markdown is the source of truth; the+compressed index is an accelerator built on top of it — never instead of it. That ordering is the whole+trick: <strong>compression buys speed; the preserved original buys correctness.</strong> You get the small wake-up+digest without ever betting your only copy on a summary.</p>+<h2 id="when-not-to-compress" tabindex="-1">When not to compress <a class="anchor" href="#when-not-to-compress" aria-hidden="true">#</a></h2>+<p>Compression is a tradeoff, not a default. Skip it (or always keep the original alongside) when the working+set already fits comfortably — compressing tiny context adds latency and lossy risk for no gain; when the+task is <strong>audit- or compliance-critical</strong> and you need the exact source; when the content is full of+<strong>exceptions and edge-cases</strong> a summary tends to flatten; or when data carries <strong>classification or+provenance</strong> that must travel with it. The cross-cutting rule again: never let the compressed copy become+the <em>only</em> copy of something you might need to defend or trace.</p>+<h2 id="faq" tabindex="-1">FAQ <a class="anchor" href="#faq" aria-hidden="true">#</a></h2>+<p><strong>Is this the same as RAG?</strong> Related but not the same. Retrieval fetches relevant chunks; compression+changes how densely those chunks are stored and loaded. They compose — you retrieve, and what you retrieve+can be compressed for a smaller, sharper context.</p>+<p><strong>How much can I compress safely?</strong> As much as you like <em>if</em> the lossless original is preserved and+pointer-linked. The compressed form’s job is fast recall; correctness lives in the untouched source you can+always fall back to.</p>+<p><strong>Does compression help cost or just context size?</strong> Both. Fewer tokens loaded per session is directly+fewer tokens billed — compression is one of the cheapest levers on an agent’s running cost.</p>+<hr>+<p>Munder Difflin gives every Claude Code agent markdown memory plus+<a href="https://munderdiffl.in/#how">a shared semantic palace that keeps the lossless original and a compact, recall-fast copy</a>+— so months of context load in a few thousand tokens, locally.+<a href="https://munderdiffl.in/#install">Download Munder Difflin</a> to give your agents memory that compresses+without forgetting; it’s free and open source.</p>++++ <section class="faq" aria-label="Frequently asked questions">+ <h2 id="faq">FAQ</h2>++ <div class="qa">+ <p class="q">Why compress an agent's memory instead of using a bigger context window?</p>+ <p class="a">Because a bigger window carries more noise, costs more per call, and still fills up. The 2026 consensus is to compress what the agent carries forward — fewer, denser tokens of signal — rather than chase a larger window. Compression is what lets a memory store grow for months while the wake-up context an agent loads stays small.</p>+ </div>++ <div class="qa">+ <p class="q">Doesn't compressing memory lose information?</p>+ <p class="a">It can — compression is non-deterministically lossy, and a summary can silently drop the one exception or classification tag that mattered. The fix is to never make the compressed copy the only copy: keep the lossless original on disk and treat the compressed form as a fast-recall accelerator built from it, not a replacement.</p>+ </div>++ <div class="qa">+ <p class="q">What actually gets compressed — the stored memory or the prompt?</p>+ <p class="a">Both are valid, at different layers. You can compact the live conversation as it approaches the window limit, and you can compress long-term stored memory so a session can absorb months of history in a few thousand tokens at wake-up. This post is mostly about the second: durable memory you recall across sessions.</p>+ </div>++ </section>++++ <hr />+ <div class="share">+ <span class="lbl">Tags</span>+ <a class="tag" href="/blog/tags/internals/">Internals</a><a class="tag" href="/blog/tags/memory/">Memory</a><a class="tag" href="/blog/tags/mempalace/">MemPalace</a><a class="tag" href="/blog/tags/multi-agent/">Multi-Agent</a>+ </div>++ </div>+++ <aside class="toc-rail" aria-label="Table of contents">+ <div class="tt">On this page</div>+ <ol>+ <li class="lvl-2"><a href="#why-compress-at-all">Why compress at all</a></li><li class="lvl-2"><a href="#the-compression-toolbox">The compression toolbox</a></li><li class="lvl-2"><a href="#the-trap-lossy-in-ways-you-cant-predict">The trap: lossy in ways you can’t predict</a></li><li class="lvl-2"><a href="#the-pattern-keep-the-original-compress-a-copy">The pattern: keep the original, compress a copy</a></li><li class="lvl-2"><a href="#when-not-to-compress">When not to compress</a></li><li class="lvl-2"><a href="#faq">FAQ</a></li>+ </ol>+ </aside>++ </div>++ <footer class="post-foot wrap"><div class="prevnext">+ <a class="pn prev" href="/blog/generative-engine-optimization-techniques/"><span class="k">← Older</span><span class="t">Generative Engine Optimization: The Techniques the Research Backs</span></a>+ <a class="pn next" href="/blog/cline-vs-munder-difflin/"><span class="k">Newer →</span><span class="t">Cline vs Munder Difflin: In-Editor Agent or Local Hive?</span></a>+ </div>+ <section class="related">+ <h2>Related in Internals</h2>+ <div class="post-grid">+ <article class="card">+ <a class="thumb t-internals" href="/blog/architecture-two-planes-one-renderer/" aria-hidden="true" tabindex="-1">+ <span>INTERNALS</span>+ </a>+ <div class="body">+ <div class="meta">+ <time datetime="2026-06-04">Jun 4, 2026</time>+ <span class="dot" aria-hidden="true"></span>+ <span>5 min</span>+ </div>+ <h3><a href="/blog/architecture-two-planes-one-renderer/">Munder Difflin's Architecture: Two Data Planes, One Renderer</a></h3>+ <p class="dek">A walkthrough of the multi-agent harness architecture: a node-pty terminal plane and a hooks/hive event plane feeding one React + Pixi.js renderer.</p>+ <div class="foot">+ <span class="tag kind">Internals</span>+ <a class="more" href="/blog/architecture-two-planes-one-renderer/" aria-label="Read: Munder Difflin's Architecture: Two Data Planes, One Renderer">Read →</a>+ </div>+ </div>+</article>+<article class="card">+ <a class="thumb t-internals" href="/blog/file-based-coordination-vs-message-queues/" aria-hidden="true" tabindex="-1">+ <span>INTERNALS</span>+ </a>+ <div class="body">+ <div class="meta">+ <time datetime="2026-06-04">Jun 4, 2026</time>+ <span class="dot" aria-hidden="true"></span>+ <span>4 min</span>+ </div>+ <h3><a href="/blog/file-based-coordination-vs-message-queues/">File-Based Coordination vs Message Queues for AI Agents</a></h3>+ <p class="dek">Why a local agent hive coordinates through plain files, not Redis or RabbitMQ: the zero-ops wins, the real tradeoffs, and where files hit a ceiling.</p>+ <div class="foot">+ <span class="tag kind">Internals</span>+ <a class="more" href="/blog/file-based-coordination-vs-message-queues/" aria-label="Read: File-Based Coordination vs Message Queues for AI Agents">Read →</a>+ </div>+ </div>+</article>+<article class="card">+ <a class="thumb t-internals" href="/blog/observability-for-agent-fleets/" aria-hidden="true" tabindex="-1">+ <span>INTERNALS</span>+ </a>+ <div class="body">+ <div class="meta">+ <time datetime="2026-06-04">Jun 4, 2026</time>+ <span class="dot" aria-hidden="true"></span>+ <span>4 min</span>+ </div>+ <h3><a href="/blog/observability-for-agent-fleets/">Observability for Agent Fleets: Seeing What Your Agents Do</a></h3>+ <p class="dek">Agent fleet observability: the four questions your dashboard must answer about who's working, what they're doing, what it costs, and what they know.</p>+ <div class="foot">+ <span class="tag kind">Internals</span>+ <a class="more" href="/blog/observability-for-agent-fleets/" aria-label="Read: Observability for Agent Fleets: Seeing What Your Agents Do">Read →</a>+ </div>+ </div>+</article>++ </div>+ </section>++ </footer>+</article>++</main>+<footer>+ <div class="wrap">+ <div class="foot-top">+ <a class="foot-brand brand" href="/blog/">+ <span class="brand-mark">MD</span>+ <span class="brand-word">MUNDER DIFFLIN<span class="sep">/</span><span class="blog">BLOG</span></span>+ </a>+ <div class="foot-links">+ <a href="/blog/">Latest</a>+ <a href="/blog/topics/">Topics</a>+ <a href="/blog/about/">About</a>+ <a href="https://munderdiffl.in/blog/feed.xml">RSS</a>+ <a href="https://github.com/chaitanyagiri/munder-difflin">GitHub</a>+ <a href="https://munderdiffl.in">munderdiffl.in ↗</a>+ </div>+ </div>+ <div class="foot-meta">+ © 2026 Munder Difflin · Local-first · MIT-licensed ·+ Built on <a href="https://munderdiffl.in">Electron, React, Pixi.js, xterm.js & node-pty</a>.+ Run an office of agents while you sleep.+ </div>+ </div>+</footer>++<script>+ // sticky-nav border on scroll+ (function () {+ var nav = document.getElementById('nav');+ if (!nav) return;+ var onScroll = function () { nav.classList.toggle('scrolled', window.scrollY > 8); };+ onScroll(); window.addEventListener('scroll', onScroll, { passive: true });+ })();+</script>++</body>+</html>diff --git a/docs/blog/feed.xml b/docs/blog/feed.xmlindex bbcf2775..51caaa19 100644--- a/docs/blog/feed.xml+++ b/docs/blog/feed.xml@@ -15,6 +15,14 @@<updated>2026-06-05T00:00:00.000Z</updated><category term="Comparisons" /><summary>A fair comparison of Cline and Munder Difflin — an in-editor BYOK coding agent vs a local multi-agent orchestration hive — and when to pick which.</summary></entry>+ <entry>+ <title>Compressing Agent Memory Without Losing the Original</title>+ <link href="https://munderdiffl.in/blog/compressing-agent-memory/" />+ <id>https://munderdiffl.in/blog/compressing-agent-memory/</id>+ <published>2026-06-05T00:00:00.000Z</published>+ <updated>2026-06-05T00:00:00.000Z</updated><category term="Internals" />+ <summary>How and why to compress an agent's long-term memory: the toolbox, the lossy trap, and keeping a lossless original beside a compact copy.</summary>+ </entry><entry><title>Generative Engine Optimization: The Techniques the Research Backs</title><link href="https://munderdiffl.in/blog/generative-engine-optimization-techniques/" />@@ -479,12 +487,4 @@<updated>2026-06-02T00:00:00.000Z</updated><category term="Use Cases" /><summary>What it takes to run a self-coordinating office of AI coding agents that keeps shipping after you log off — and the guardrails that keep it sane.</summary></entry>- <entry>- <title>The Single-Committer Pattern: Multi-Agent Git Without Corruption</title>- <link href="https://munderdiffl.in/blog/single-committer-git-pattern/" />- <id>https://munderdiffl.in/blog/single-committer-git-pattern/</id>- <published>2026-06-02T00:00:00.000Z</published>- <updated>2026-06-02T00:00:00.000Z</updated><category term="Internals" />- <summary>Parallel agents corrupt a repo with index.lock races. The single-committer pattern — agents write, one process commits — fixes concurrent git writes.</summary>- </entry></feed>diff --git a/docs/blog/file-based-coordination-vs-message-queues/index.html b/docs/blog/file-based-coordination-vs-message-queues/index.htmlindex b0fd655b..2cfca2b6 100644--- a/docs/blog/file-based-coordination-vs-message-queues/index.html+++ b/docs/blog/file-based-coordination-vs-message-queues/index.html@@ -262,56 +262,56 @@ durable, git-versioned, and debuggable by just looking.<h2>Related in Internals</h2><div class="post-grid"><article class="card">- <a class="thumb t-internals" href="/blog/architecture-two-planes-one-renderer/" aria-hidden="true" tabindex="-1">+ <a class="thumb t-internals" href="/blog/compressing-agent-memory/" aria-hidden="true" tabindex="-1"><span>INTERNALS</span></a><div class="body"><div class="meta">- <time datetime="2026-06-04">Jun 4, 2026</time>+ <time datetime="2026-06-05">Jun 5, 2026</time><span class="dot" aria-hidden="true"></span><span>5 min</span></div>- <h3><a href="/blog/architecture-two-planes-one-renderer/">Munder Difflin's Architecture: Two Data Planes, One Renderer</a></h3>- <p class="dek">A walkthrough of the multi-agent harness architecture: a node-pty terminal plane and a hooks/hive event plane feeding one React + Pixi.js renderer.</p>+ <h3><a href="/blog/compressing-agent-memory/">Compressing Agent Memory Without Losing the Original</a></h3>+ <p class="dek">How and why to compress an agent's long-term memory: the toolbox, the lossy trap, and keeping a lossless original beside a compact copy.</p><div class="foot"><span class="tag kind">Internals</span>- <a class="more" href="/blog/architecture-two-planes-one-renderer/" aria-label="Read: Munder Difflin's Architecture: Two Data Planes, One Renderer">Read →</a>+ <a class="more" href="/blog/compressing-agent-memory/" aria-label="Read: Compressing Agent Memory Without Losing the Original">Read →</a></div></div></article><article class="card">- <a class="thumb t-internals" href="/blog/observability-for-agent-fleets/" aria-hidden="true" tabindex="-1">+ <a class="thumb t-internals" href="/blog/architecture-two-planes-one-renderer/" aria-hidden="true" tabindex="-1"><span>INTERNALS</span></a><div class="body"><div class="meta"><time datetime="2026-06-04">Jun 4, 2026</time><span class="dot" aria-hidden="true"></span>- <span>4 min</span>+ <span>5 min</span></div>- <h3><a href="/blog/observability-for-agent-fleets/">Observability for Agent Fleets: Seeing What Your Agents Do</a></h3>- <p class="dek">Agent fleet observability: the four questions your dashboard must answer about who's working, what they're doing, what it costs, and what they know.</p>+ <h3><a href="/blog/architecture-two-planes-one-renderer/">Munder Difflin's Architecture: Two Data Planes, One Renderer</a></h3>+ <p class="dek">A walkthrough of the multi-agent harness architecture: a node-pty terminal plane and a hooks/hive event plane feeding one React + Pixi.js renderer.</p><div class="foot"><span class="tag kind">Internals</span>- <a class="more" href="/blog/observability-for-agent-fleets/" aria-label="Read: Observability for Agent Fleets: Seeing What Your Agents Do">Read →</a>+ <a class="more" href="/blog/architecture-two-planes-one-renderer/" aria-label="Read: Munder Difflin's Architecture: Two Data Planes, One Renderer">Read →</a></div></div></article><article class="card">- <a class="thumb t-internals" href="/blog/recovering-from-agent-failures/" aria-hidden="true" tabindex="-1">+ <a class="thumb t-internals" href="/blog/observability-for-agent-fleets/" aria-hidden="true" tabindex="-1"><span>INTERNALS</span></a><div class="body"><div class="meta"><time datetime="2026-06-04">Jun 4, 2026</time><span class="dot" aria-hidden="true"></span>- <span>6 min</span>+ <span>4 min</span></div>- <h3><a href="/blog/recovering-from-agent-failures/">Recovering From Agent Failures: Resilience Patterns for a Hive</a></h3>- <p class="dek">How multi-agent systems survive partial failure: retry with stale-lock recovery, poison-message quarantine, hop-cap circuit breaking, idempotent handling.</p>+ <h3><a href="/blog/observability-for-agent-fleets/">Observability for Agent Fleets: Seeing What Your Agents Do</a></h3>+ <p class="dek">Agent fleet observability: the four questions your dashboard must answer about who's working, what they're doing, what it costs, and what they know.</p><div class="foot"><span class="tag kind">Internals</span>- <a class="more" href="/blog/recovering-from-agent-failures/" aria-label="Read: Recovering From Agent Failures: Resilience Patterns for a Hive">Read →</a>+ <a class="more" href="/blog/observability-for-agent-fleets/" aria-label="Read: Observability for Agent Fleets: Seeing What Your Agents Do">Read →</a></div></div></article>diff --git a/docs/blog/generative-engine-optimization-techniques/index.html b/docs/blog/generative-engine-optimization-techniques/index.htmlindex 711898fb..a220a37a 100644--- a/docs/blog/generative-engine-optimization-techniques/index.html+++ b/docs/blog/generative-engine-optimization-techniques/index.html@@ -274,7 +274,7 @@ and quotable. <a href="/#install">Download Munder Difflin</a> to see it; free an<footer class="post-foot wrap"><div class="prevnext"><a class="pn prev" href="/blog/how-to-install-and-use-munder-difflin/"><span class="k">← Older</span><span class="t">How to Install and Use Munder Difflin</span></a>- <a class="pn next" href="/blog/cline-vs-munder-difflin/"><span class="k">Newer →</span><span class="t">Cline vs Munder Difflin: In-Editor Agent or Local Hive?</span></a>+ <a class="pn next" href="/blog/compressing-agent-memory/"><span class="k">Newer →</span><span class="t">Compressing Agent Memory Without Losing the Original</span></a></div><section class="related"><h2>Related in Concepts</h2>diff --git a/docs/blog/index.html b/docs/blog/index.htmlindex 5ca72404..39588ca4 100644--- a/docs/blog/index.html+++ b/docs/blog/index.html@@ -85,11 +85,11 @@building a self-coordinating hive of agents with a GOD orchestrator you talk to.</p><div class="filters" role="navigation" aria-label="Topics">- <a class="chip" aria-current="page">All<span class="ct">77</span></a>+ <a class="chip" aria-current="page">All<span class="ct">78</span></a><a class="chip" href="/blog/topics/guides/">guides<span class="ct">21</span></a>- <a class="chip" href="/blog/topics/internals/">internals<span class="ct">14</span></a>+ <a class="chip" href="/blog/topics/internals/">internals<span class="ct">15</span></a><a class="chip" href="/blog/topics/concepts/">concepts<span class="ct">13</span></a>@@ -132,6 +132,24 @@<section class="wrap"><div class="post-grid"><article class="card">+ <a class="thumb t-internals" href="/blog/compressing-agent-memory/" aria-hidden="true" tabindex="-1">+ <span>INTERNALS</span>+ </a>+ <div class="body">+ <div class="meta">+ <time datetime="2026-06-05">Jun 5, 2026</time>+ <span class="dot" aria-hidden="true"></span>+ <span>5 min</span>+ </div>+ <h3><a href="/blog/compressing-agent-memory/">Compressing Agent Memory Without Losing the Original</a></h3>+ <p class="dek">How and why to compress an agent's long-term memory: the toolbox, the lossy trap, and keeping a lossless original beside a compact copy.</p>+ <div class="foot">+ <span class="tag kind">Internals</span>+ <a class="more" href="/blog/compressing-agent-memory/" aria-label="Read: Compressing Agent Memory Without Losing the Original">Read →</a>+ </div>+ </div>+</article>+<article class="card"><a class="thumb t-concepts" href="/blog/generative-engine-optimization-techniques/" aria-hidden="true" tabindex="-1"><span>CONCEPTS</span></a>diff --git a/docs/blog/node-pty-electron-real-terminals/index.html b/docs/blog/node-pty-electron-real-terminals/index.htmlindex 718e94ee..d7759a4c 100644--- a/docs/blog/node-pty-electron-real-terminals/index.html+++ b/docs/blog/node-pty-electron-real-terminals/index.html@@ -324,43 +324,43 @@ real terminals driving real agents; it’s free and open source.</p><h2>Related in Internals</h2><div class="post-grid"><article class="card">- <a class="thumb t-internals" href="/blog/architecture-two-planes-one-renderer/" aria-hidden="true" tabindex="-1">+ <a class="thumb t-internals" href="/blog/compressing-agent-memory/" aria-hidden="true" tabindex="-1"><span>INTERNALS</span></a><div class="body"><div class="meta">- <time datetime="2026-06-04">Jun 4, 2026</time>+ <time datetime="2026-06-05">Jun 5, 2026</time><span class="dot" aria-hidden="true"></span><span>5 min</span></div>- <h3><a href="/blog/architecture-two-planes-one-renderer/">Munder Difflin's Architecture: Two Data Planes, One Renderer</a></h3>- <p class="dek">A walkthrough of the multi-agent harness architecture: a node-pty terminal plane and a hooks/hive event plane feeding one React + Pixi.js renderer.</p>+ <h3><a href="/blog/compressing-agent-memory/">Compressing Agent Memory Without Losing the Original</a></h3>+ <p class="dek">How and why to compress an agent's long-term memory: the toolbox, the lossy trap, and keeping a lossless original beside a compact copy.</p><div class="foot"><span class="tag kind">Internals</span>- <a class="more" href="/blog/architecture-two-planes-one-renderer/" aria-label="Read: Munder Difflin's Architecture: Two Data Planes, One Renderer">Read →</a>+ <a class="more" href="/blog/compressing-agent-memory/" aria-label="Read: Compressing Agent Memory Without Losing the Original">Read →</a></div></div></article><article class="card">- <a class="thumb t-internals" href="/blog/file-based-coordination-vs-message-queues/" aria-hidden="true" tabindex="-1">+ <a class="thumb t-internals" href="/blog/architecture-two-planes-one-renderer/" aria-hidden="true" tabindex="-1"><span>INTERNALS</span></a><div class="body"><div class="meta"><time datetime="2026-06-04">Jun 4, 2026</time><span class="dot" aria-hidden="true"></span>- <span>4 min</span>+ <span>5 min</span></div>- <h3><a href="/blog/file-based-coordination-vs-message-queues/">File-Based Coordination vs Message Queues for AI Agents</a></h3>- <p class="dek">Why a local agent hive coordinates through plain files, not Redis or RabbitMQ: the zero-ops wins, the real tradeoffs, and where files hit a ceiling.</p>+ <h3><a href="/blog/architecture-two-planes-one-renderer/">Munder Difflin's Architecture: Two Data Planes, One Renderer</a></h3>+ <p class="dek">A walkthrough of the multi-agent harness architecture: a node-pty terminal plane and a hooks/hive event plane feeding one React + Pixi.js renderer.</p><div class="foot"><span class="tag kind">Internals</span>- <a class="more" href="/blog/file-based-coordination-vs-message-queues/" aria-label="Read: File-Based Coordination vs Message Queues for AI Agents">Read →</a>+ <a class="more" href="/blog/architecture-two-planes-one-renderer/" aria-label="Read: Munder Difflin's Architecture: Two Data Planes, One Renderer">Read →</a></div></div></article><article class="card">- <a class="thumb t-internals" href="/blog/observability-for-agent-fleets/" aria-hidden="true" tabindex="-1">+ <a class="thumb t-internals" href="/blog/file-based-coordination-vs-message-queues/" aria-hidden="true" tabindex="-1"><span>INTERNALS</span></a><div class="body">@@ -369,11 +369,11 @@ real terminals driving real agents; it’s free and open source.</p><span class="dot" aria-hidden="true"></span><span>4 min</span></div>- <h3><a href="/blog/observability-for-agent-fleets/">Observability for Agent Fleets: Seeing What Your Agents Do</a></h3>- <p class="dek">Agent fleet observability: the four questions your dashboard must answer about who's working, what they're doing, what it costs, and what they know.</p>+ <h3><a href="/blog/file-based-coordination-vs-message-queues/">File-Based Coordination vs Message Queues for AI Agents</a></h3>+ <p class="dek">Why a local agent hive coordinates through plain files, not Redis or RabbitMQ: the zero-ops wins, the real tradeoffs, and where files hit a ceiling.</p><div class="foot"><span class="tag kind">Internals</span>- <a class="more" href="/blog/observability-for-agent-fleets/" aria-label="Read: Observability for Agent Fleets: Seeing What Your Agents Do">Read →</a>+ <a class="more" href="/blog/file-based-coordination-vs-message-queues/" aria-label="Read: File-Based Coordination vs Message Queues for AI Agents">Read →</a></div></div></article>diff --git a/docs/blog/observability-for-agent-fleets/index.html b/docs/blog/observability-for-agent-fleets/index.htmlindex ee7835eb..7ac73723 100644--- a/docs/blog/observability-for-agent-fleets/index.html+++ b/docs/blog/observability-for-agent-fleets/index.html@@ -261,56 +261,56 @@ open source.</p><h2>Related in Internals</h2><div class="post-grid"><article class="card">- <a class="thumb t-internals" href="/blog/architecture-two-planes-one-renderer/" aria-hidden="true" tabindex="-1">+ <a class="thumb t-internals" href="/blog/compressing-agent-memory/" aria-hidden="true" tabindex="-1"><span>INTERNALS</span></a><div class="body"><div class="meta">- <time datetime="2026-06-04">Jun 4, 2026</time>+ <time datetime="2026-06-05">Jun 5, 2026</time><span class="dot" aria-hidden="true"></span><span>5 min</span></div>- <h3><a href="/blog/architecture-two-planes-one-renderer/">Munder Difflin's Architecture: Two Data Planes, One Renderer</a></h3>- <p class="dek">A walkthrough of the multi-agent harness architecture: a node-pty terminal plane and a hooks/hive event plane feeding one React + Pixi.js renderer.</p>+ <h3><a href="/blog/compressing-agent-memory/">Compressing Agent Memory Without Losing the Original</a></h3>+ <p class="dek">How and why to compress an agent's long-term memory: the toolbox, the lossy trap, and keeping a lossless original beside a compact copy.</p><div class="foot"><span class="tag kind">Internals</span>- <a class="more" href="/blog/architecture-two-planes-one-renderer/" aria-label="Read: Munder Difflin's Architecture: Two Data Planes, One Renderer">Read →</a>+ <a class="more" href="/blog/compressing-agent-memory/" aria-label="Read: Compressing Agent Memory Without Losing the Original">Read →</a></div></div></article><article class="card">- <a class="thumb t-internals" href="/blog/file-based-coordination-vs-message-queues/" aria-hidden="true" tabindex="-1">+ <a class="thumb t-internals" href="/blog/architecture-two-planes-one-renderer/" aria-hidden="true" tabindex="-1"><span>INTERNALS</span></a><div class="body"><div class="meta"><time datetime="2026-06-04">Jun 4, 2026</time><span class="dot" aria-hidden="true"></span>- <span>4 min</span>+ <span>5 min</span></div>- <h3><a href="/blog/file-based-coordination-vs-message-queues/">File-Based Coordination vs Message Queues for AI Agents</a></h3>- <p class="dek">Why a local agent hive coordinates through plain files, not Redis or RabbitMQ: the zero-ops wins, the real tradeoffs, and where files hit a ceiling.</p>+ <h3><a href="/blog/architecture-two-planes-one-renderer/">Munder Difflin's Architecture: Two Data Planes, One Renderer</a></h3>+ <p class="dek">A walkthrough of the multi-agent harness architecture: a node-pty terminal plane and a hooks/hive event plane feeding one React + Pixi.js renderer.</p><div class="foot"><span class="tag kind">Internals</span>- <a class="more" href="/blog/file-based-coordination-vs-message-queues/" aria-label="Read: File-Based Coordination vs Message Queues for AI Agents">Read →</a>+ <a class="more" href="/blog/architecture-two-planes-one-renderer/" aria-label="Read: Munder Difflin's Architecture: Two Data Planes, One Renderer">Read →</a></div></div></article><article class="card">- <a class="thumb t-internals" href="/blog/recovering-from-agent-failures/" aria-hidden="true" tabindex="-1">+ <a class="thumb t-internals" href="/blog/file-based-coordination-vs-message-queues/" aria-hidden="true" tabindex="-1"><span>INTERNALS</span></a><div class="body"><div class="meta"><time datetime="2026-06-04">Jun 4, 2026</time><span class="dot" aria-hidden="true"></span>- <span>6 min</span>+ <span>4 min</span></div>- <h3><a href="/blog/recovering-from-agent-failures/">Recovering From Agent Failures: Resilience Patterns for a Hive</a></h3>- <p class="dek">How multi-agent systems survive partial failure: retry with stale-lock recovery, poison-message quarantine, hop-cap circuit breaking, idempotent handling.</p>+ <h3><a href="/blog/file-based-coordination-vs-message-queues/">File-Based Coordination vs Message Queues for AI Agents</a></h3>+ <p class="dek">Why a local agent hive coordinates through plain files, not Redis or RabbitMQ: the zero-ops wins, the real tradeoffs, and where files hit a ceiling.</p><div class="foot"><span class="tag kind">Internals</span>- <a class="more" href="/blog/recovering-from-agent-failures/" aria-label="Read: Recovering From Agent Failures: Resilience Patterns for a Hive">Read →</a>+ <a class="more" href="/blog/file-based-coordination-vs-message-queues/" aria-label="Read: File-Based Coordination vs Message Queues for AI Agents">Read →</a></div></div></article>diff --git a/docs/blog/recovering-from-agent-failures/index.html b/docs/blog/recovering-from-agent-failures/index.htmlindex ea3b3845..d1648d77 100644--- a/docs/blog/recovering-from-agent-failures/index.html+++ b/docs/blog/recovering-from-agent-failures/index.html@@ -281,43 +281,43 @@ there to read.</p><h2>Related in Internals</h2><div class="post-grid"><article class="card">- <a class="thumb t-internals" href="/blog/architecture-two-planes-one-renderer/" aria-hidden="true" tabindex="-1">+ <a class="thumb t-internals" href="/blog/compressing-agent-memory/" aria-hidden="true" tabindex="-1"><span>INTERNALS</span></a><div class="body"><div class="meta">- <time datetime="2026-06-04">Jun 4, 2026</time>+ <time datetime="2026-06-05">Jun 5, 2026</time><span class="dot" aria-hidden="true"></span><span>5 min</span></div>- <h3><a href="/blog/architecture-two-planes-one-renderer/">Munder Difflin's Architecture: Two Data Planes, One Renderer</a></h3>- <p class="dek">A walkthrough of the multi-agent harness architecture: a node-pty terminal plane and a hooks/hive event plane feeding one React + Pixi.js renderer.</p>+ <h3><a href="/blog/compressing-agent-memory/">Compressing Agent Memory Without Losing the Original</a></h3>+ <p class="dek">How and why to compress an agent's long-term memory: the toolbox, the lossy trap, and keeping a lossless original beside a compact copy.</p><div class="foot"><span class="tag kind">Internals</span>- <a class="more" href="/blog/architecture-two-planes-one-renderer/" aria-label="Read: Munder Difflin's Architecture: Two Data Planes, One Renderer">Read →</a>+ <a class="more" href="/blog/compressing-agent-memory/" aria-label="Read: Compressing Agent Memory Without Losing the Original">Read →</a></div></div></article><article class="card">- <a class="thumb t-internals" href="/blog/file-based-coordination-vs-message-queues/" aria-hidden="true" tabindex="-1">+ <a class="thumb t-internals" href="/blog/architecture-two-planes-one-renderer/" aria-hidden="true" tabindex="-1"><span>INTERNALS</span></a><div class="body"><div class="meta"><time datetime="2026-06-04">Jun 4, 2026</time><span class="dot" aria-hidden="true"></span>- <span>4 min</span>+ <span>5 min</span></div>- <h3><a href="/blog/file-based-coordination-vs-message-queues/">File-Based Coordination vs Message Queues for AI Agents</a></h3>- <p class="dek">Why a local agent hive coordinates through plain files, not Redis or RabbitMQ: the zero-ops wins, the real tradeoffs, and where files hit a ceiling.</p>+ <h3><a href="/blog/architecture-two-planes-one-renderer/">Munder Difflin's Architecture: Two Data Planes, One Renderer</a></h3>+ <p class="dek">A walkthrough of the multi-agent harness architecture: a node-pty terminal plane and a hooks/hive event plane feeding one React + Pixi.js renderer.</p><div class="foot"><span class="tag kind">Internals</span>- <a class="more" href="/blog/file-based-coordination-vs-message-queues/" aria-label="Read: File-Based Coordination vs Message Queues for AI Agents">Read →</a>+ <a class="more" href="/blog/architecture-two-planes-one-renderer/" aria-label="Read: Munder Difflin's Architecture: Two Data Planes, One Renderer">Read →</a></div></div></article><article class="card">- <a class="thumb t-internals" href="/blog/observability-for-agent-fleets/" aria-hidden="true" tabindex="-1">+ <a class="thumb t-internals" href="/blog/file-based-coordination-vs-message-queues/" aria-hidden="true" tabindex="-1"><span>INTERNALS</span></a><div class="body">@@ -326,11 +326,11 @@ there to read.</p><span class="dot" aria-hidden="true"></span><span>4 min</span></div>- <h3><a href="/blog/observability-for-agent-fleets/">Observability for Agent Fleets: Seeing What Your Agents Do</a></h3>- <p class="dek">Agent fleet observability: the four questions your dashboard must answer about who's working, what they're doing, what it costs, and what they know.</p>+ <h3><a href="/blog/file-based-coordination-vs-message-queues/">File-Based Coordination vs Message Queues for AI Agents</a></h3>+ <p class="dek">Why a local agent hive coordinates through plain files, not Redis or RabbitMQ: the zero-ops wins, the real tradeoffs, and where files hit a ceiling.</p><div class="foot"><span class="tag kind">Internals</span>- <a class="more" href="/blog/observability-for-agent-fleets/" aria-label="Read: Observability for Agent Fleets: Seeing What Your Agents Do">Read →</a>+ <a class="more" href="/blog/file-based-coordination-vs-message-queues/" aria-label="Read: File-Based Coordination vs Message Queues for AI Agents">Read →</a></div></div></article>diff --git a/docs/blog/rendering-many-live-terminals-performance/index.html b/docs/blog/rendering-many-live-terminals-performance/index.htmlindex fc779cbd..330fc7c1 100644--- a/docs/blog/rendering-many-live-terminals-performance/index.html+++ b/docs/blog/rendering-many-live-terminals-performance/index.html@@ -284,43 +284,43 @@ to run many real terminals without melting your CPU; it’s free and open source<h2>Related in Internals</h2><div class="post-grid"><article class="card">- <a class="thumb t-internals" href="/blog/architecture-two-planes-one-renderer/" aria-hidden="true" tabindex="-1">+ <a class="thumb t-internals" href="/blog/compressing-agent-memory/" aria-hidden="true" tabindex="-1"><span>INTERNALS</span></a><div class="body"><div class="meta">- <time datetime="2026-06-04">Jun 4, 2026</time>+ <time datetime="2026-06-05">Jun 5, 2026</time><span class="dot" aria-hidden="true"></span><span>5 min</span></div>- <h3><a href="/blog/architecture-two-planes-one-renderer/">Munder Difflin's Architecture: Two Data Planes, One Renderer</a></h3>- <p class="dek">A walkthrough of the multi-agent harness architecture: a node-pty terminal plane and a hooks/hive event plane feeding one React + Pixi.js renderer.</p>+ <h3><a href="/blog/compressing-agent-memory/">Compressing Agent Memory Without Losing the Original</a></h3>+ <p class="dek">How and why to compress an agent's long-term memory: the toolbox, the lossy trap, and keeping a lossless original beside a compact copy.</p><div class="foot"><span class="tag kind">Internals</span>- <a class="more" href="/blog/architecture-two-planes-one-renderer/" aria-label="Read: Munder Difflin's Architecture: Two Data Planes, One Renderer">Read →</a>+ <a class="more" href="/blog/compressing-agent-memory/" aria-label="Read: Compressing Agent Memory Without Losing the Original">Read →</a></div></div></article><article class="card">- <a class="thumb t-internals" href="/blog/file-based-coordination-vs-message-queues/" aria-hidden="true" tabindex="-1">+ <a class="thumb t-internals" href="/blog/architecture-two-planes-one-renderer/" aria-hidden="true" tabindex="-1"><span>INTERNALS</span></a><div class="body"><div class="meta"><time datetime="2026-06-04">Jun 4, 2026</time><span class="dot" aria-hidden="true"></span>- <span>4 min</span>+ <span>5 min</span></div>- <h3><a href="/blog/file-based-coordination-vs-message-queues/">File-Based Coordination vs Message Queues for AI Agents</a></h3>- <p class="dek">Why a local agent hive coordinates through plain files, not Redis or RabbitMQ: the zero-ops wins, the real tradeoffs, and where files hit a ceiling.</p>+ <h3><a href="/blog/architecture-two-planes-one-renderer/">Munder Difflin's Architecture: Two Data Planes, One Renderer</a></h3>+ <p class="dek">A walkthrough of the multi-agent harness architecture: a node-pty terminal plane and a hooks/hive event plane feeding one React + Pixi.js renderer.</p><div class="foot"><span class="tag kind">Internals</span>- <a class="more" href="/blog/file-based-coordination-vs-message-queues/" aria-label="Read: File-Based Coordination vs Message Queues for AI Agents">Read →</a>+ <a class="more" href="/blog/architecture-two-planes-one-renderer/" aria-label="Read: Munder Difflin's Architecture: Two Data Planes, One Renderer">Read →</a></div></div></article><article class="card">- <a class="thumb t-internals" href="/blog/observability-for-agent-fleets/" aria-hidden="true" tabindex="-1">+ <a class="thumb t-internals" href="/blog/file-based-coordination-vs-message-queues/" aria-hidden="true" tabindex="-1"><span>INTERNALS</span></a><div class="body">@@ -329,11 +329,11 @@ to run many real terminals without melting your CPU; it’s free and open source<span class="dot" aria-hidden="true"></span><span>4 min</span></div>- <h3><a href="/blog/observability-for-agent-fleets/">Observability for Agent Fleets: Seeing What Your Agents Do</a></h3>- <p class="dek">Agent fleet observability: the four questions your dashboard must answer about who's working, what they're doing, what it costs, and what they know.</p>+ <h3><a href="/blog/file-based-coordination-vs-message-queues/">File-Based Coordination vs Message Queues for AI Agents</a></h3>+ <p class="dek">Why a local agent hive coordinates through plain files, not Redis or RabbitMQ: the zero-ops wins, the real tradeoffs, and where files hit a ceiling.</p><div class="foot"><span class="tag kind">Internals</span>- <a class="more" href="/blog/observability-for-agent-fleets/" aria-label="Read: Observability for Agent Fleets: Seeing What Your Agents Do">Read →</a>+ <a class="more" href="/blog/file-based-coordination-vs-message-queues/" aria-label="Read: File-Based Coordination vs Message Queues for AI Agents">Read →</a></div></div></article>diff --git a/docs/blog/single-committer-git-pattern/index.html b/docs/blog/single-committer-git-pattern/index.htmlindex f0709d5e..49ff3d38 100644--- a/docs/blog/single-committer-git-pattern/index.html+++ b/docs/blog/single-committer-git-pattern/index.html@@ -291,43 +291,43 @@ never corrupt their own state; it’s free and open source.</p><h2>Related in Internals</h2><div class="post-grid"><article class="card">- <a class="thumb t-internals" href="/blog/architecture-two-planes-one-renderer/" aria-hidden="true" tabindex="-1">+ <a class="thumb t-internals" href="/blog/compressing-agent-memory/" aria-hidden="true" tabindex="-1"><span>INTERNALS</span></a><div class="body"><div class="meta">- <time datetime="2026-06-04">Jun 4, 2026</time>+ <time datetime="2026-06-05">Jun 5, 2026</time><span class="dot" aria-hidden="true"></span><span>5 min</span></div>- <h3><a href="/blog/architecture-two-planes-one-renderer/">Munder Difflin's Architecture: Two Data Planes, One Renderer</a></h3>- <p class="dek">A walkthrough of the multi-agent harness architecture: a node-pty terminal plane and a hooks/hive event plane feeding one React + Pixi.js renderer.</p>+ <h3><a href="/blog/compressing-agent-memory/">Compressing Agent Memory Without Losing the Original</a></h3>+ <p class="dek">How and why to compress an agent's long-term memory: the toolbox, the lossy trap, and keeping a lossless original beside a compact copy.</p><div class="foot"><span class="tag kind">Internals</span>- <a class="more" href="/blog/architecture-two-planes-one-renderer/" aria-label="Read: Munder Difflin's Architecture: Two Data Planes, One Renderer">Read →</a>+ <a class="more" href="/blog/compressing-agent-memory/" aria-label="Read: Compressing Agent Memory Without Losing the Original">Read →</a></div></div></article><article class="card">- <a class="thumb t-internals" href="/blog/file-based-coordination-vs-message-queues/" aria-hidden="true" tabindex="-1">+ <a class="thumb t-internals" href="/blog/architecture-two-planes-one-renderer/" aria-hidden="true" tabindex="-1"><span>INTERNALS</span></a><div class="body"><div class="meta"><time datetime="2026-06-04">Jun 4, 2026</time><span class="dot" aria-hidden="true"></span>- <span>4 min</span>+ <span>5 min</span></div>- <h3><a href="/blog/file-based-coordination-vs-message-queues/">File-Based Coordination vs Message Queues for AI Agents</a></h3>- <p class="dek">Why a local agent hive coordinates through plain files, not Redis or RabbitMQ: the zero-ops wins, the real tradeoffs, and where files hit a ceiling.</p>+ <h3><a href="/blog/architecture-two-planes-one-renderer/">Munder Difflin's Architecture: Two Data Planes, One Renderer</a></h3>+ <p class="dek">A walkthrough of the multi-agent harness architecture: a node-pty terminal plane and a hooks/hive event plane feeding one React + Pixi.js renderer.</p><div class="foot"><span class="tag kind">Internals</span>- <a class="more" href="/blog/file-based-coordination-vs-message-queues/" aria-label="Read: File-Based Coordination vs Message Queues for AI Agents">Read →</a>+ <a class="more" href="/blog/architecture-two-planes-one-renderer/" aria-label="Read: Munder Difflin's Architecture: Two Data Planes, One Renderer">Read →</a></div></div></article><article class="card">- <a class="thumb t-internals" href="/blog/observability-for-agent-fleets/" aria-hidden="true" tabindex="-1">+ <a class="thumb t-internals" href="/blog/file-based-coordination-vs-message-queues/" aria-hidden="true" tabindex="-1"><span>INTERNALS</span></a><div class="body">@@ -336,11 +336,11 @@ never corrupt their own state; it’s free and open source.</p><span class="dot" aria-hidden="true"></span><span>4 min</span></div>- <h3><a href="/blog/observability-for-agent-fleets/">Observability for Agent Fleets: Seeing What Your Agents Do</a></h3>- <p class="dek">Agent fleet observability: the four questions your dashboard must answer about who's working, what they're doing, what it costs, and what they know.</p>+ <h3><a href="/blog/file-based-coordination-vs-message-queues/">File-Based Coordination vs Message Queues for AI Agents</a></h3>+ <p class="dek">Why a local agent hive coordinates through plain files, not Redis or RabbitMQ: the zero-ops wins, the real tradeoffs, and where files hit a ceiling.</p><div class="foot"><span class="tag kind">Internals</span>- <a class="more" href="/blog/observability-for-agent-fleets/" aria-label="Read: Observability for Agent Fleets: Seeing What Your Agents Do">Read →</a>+ <a class="more" href="/blog/file-based-coordination-vs-message-queues/" aria-label="Read: File-Based Coordination vs Message Queues for AI Agents">Read →</a></div></div></article>diff --git a/docs/blog/tags/internals/index.html b/docs/blog/tags/internals/index.htmlindex c89b63b3..a2c27936 100644--- a/docs/blog/tags/internals/index.html+++ b/docs/blog/tags/internals/index.html@@ -58,11 +58,29 @@<header class="page-head wrap"><p class="eyebrow">Tag</p><h1>#Internals</h1>- <p class="lead">20 posts tagged <strong>Internals</strong>.</p>+ <p class="lead">21 posts tagged <strong>Internals</strong>.</p></header><section class="wrap"><div class="post-grid"><article class="card">+ <a class="thumb t-internals" href="/blog/compressing-agent-memory/" aria-hidden="true" tabindex="-1">+ <span>INTERNALS</span>+ </a>+ <div class="body">+ <div class="meta">+ <time datetime="2026-06-05">Jun 5, 2026</time>+ <span class="dot" aria-hidden="true"></span>+ <span>5 min</span>+ </div>+ <h3><a href="/blog/compressing-agent-memory/">Compressing Agent Memory Without Losing the Original</a></h3>+ <p class="dek">How and why to compress an agent's long-term memory: the toolbox, the lossy trap, and keeping a lossless original beside a compact copy.</p>+ <div class="foot">+ <span class="tag kind">Internals</span>+ <a class="more" href="/blog/compressing-agent-memory/" aria-label="Read: Compressing Agent Memory Without Losing the Original">Read →</a>+ </div>+ </div>+</article>+<article class="card"><a class="thumb t-guides" href="/blog/agent-security-and-sandboxing/" aria-hidden="true" tabindex="-1"><span>GUIDES</span></a>diff --git a/docs/blog/tags/memory/index.html b/docs/blog/tags/memory/index.htmlindex 104c945c..137947ee 100644--- a/docs/blog/tags/memory/index.html+++ b/docs/blog/tags/memory/index.html@@ -58,11 +58,29 @@<header class="page-head wrap"><p class="eyebrow">Tag</p><h1>#Memory</h1>- <p class="lead">7 posts tagged <strong>Memory</strong>.</p>+ <p class="lead">8 posts tagged <strong>Memory</strong>.</p></header><section class="wrap"><div class="post-grid"><article class="card">+ <a class="thumb t-internals" href="/blog/compressing-agent-memory/" aria-hidden="true" tabindex="-1">+ <span>INTERNALS</span>+ </a>+ <div class="body">+ <div class="meta">+ <time datetime="2026-06-05">Jun 5, 2026</time>+ <span class="dot" aria-hidden="true"></span>+ <span>5 min</span>+ </div>+ <h3><a href="/blog/compressing-agent-memory/">Compressing Agent Memory Without Losing the Original</a></h3>+ <p class="dek">How and why to compress an agent's long-term memory: the toolbox, the lossy trap, and keeping a lossless original beside a compact copy.</p>+ <div class="foot">+ <span class="tag kind">Internals</span>+ <a class="more" href="/blog/compressing-agent-memory/" aria-label="Read: Compressing Agent Memory Without Losing the Original">Read →</a>+ </div>+ </div>+</article>+<article class="card"><a class="thumb t-guides" href="/blog/context-engineering-for-ai-agents/" aria-hidden="true" tabindex="-1"><span>GUIDES</span></a>diff --git a/docs/blog/tags/mempalace/index.html b/docs/blog/tags/mempalace/index.htmlindex 10c3cf15..dc020e46 100644--- a/docs/blog/tags/mempalace/index.html+++ b/docs/blog/tags/mempalace/index.html@@ -58,11 +58,29 @@<header class="page-head wrap"><p class="eyebrow">Tag</p><h1>#MemPalace</h1>- <p class="lead">5 posts tagged <strong>MemPalace</strong>.</p>+ <p class="lead">6 posts tagged <strong>MemPalace</strong>.</p></header><section class="wrap"><div class="post-grid"><article class="card">+ <a class="thumb t-internals" href="/blog/compressing-agent-memory/" aria-hidden="true" tabindex="-1">+ <span>INTERNALS</span>+ </a>+ <div class="body">+ <div class="meta">+ <time datetime="2026-06-05">Jun 5, 2026</time>+ <span class="dot" aria-hidden="true"></span>+ <span>5 min</span>+ </div>+ <h3><a href="/blog/compressing-agent-memory/">Compressing Agent Memory Without Losing the Original</a></h3>+ <p class="dek">How and why to compress an agent's long-term memory: the toolbox, the lossy trap, and keeping a lossless original beside a compact copy.</p>+ <div class="foot">+ <span class="tag kind">Internals</span>+ <a class="more" href="/blog/compressing-agent-memory/" aria-label="Read: Compressing Agent Memory Without Losing the Original">Read →</a>+ </div>+ </div>+</article>+<article class="card"><a class="thumb t-memory" href="/blog/how-agents-remember-semantic-memory/" aria-hidden="true" tabindex="-1"><span>MEMORY</span></a>diff --git a/docs/blog/tags/multi-agent/index.html b/docs/blog/tags/multi-agent/index.htmlindex 51a6af2a..130a3d45 100644--- a/docs/blog/tags/multi-agent/index.html+++ b/docs/blog/tags/multi-agent/index.html@@ -58,7 +58,7 @@<header class="page-head wrap"><p class="eyebrow">Tag</p><h1>#Multi-Agent</h1>- <p class="lead">52 posts tagged <strong>Multi-Agent</strong>.</p>+ <p class="lead">53 posts tagged <strong>Multi-Agent</strong>.</p></header><section class="wrap"><div class="post-grid">@@ -80,6 +80,24 @@</div></div></article>+<article class="card">+ <a class="thumb t-internals" href="/blog/compressing-agent-memory/" aria-hidden="true" tabindex="-1">+ <span>INTERNALS</span>+ </a>+ <div class="body">+ <div class="meta">+ <time datetime="2026-06-05">Jun 5, 2026</time>+ <span class="dot" aria-hidden="true"></span>+ <span>5 min</span>+ </div>+ <h3><a href="/blog/compressing-agent-memory/">Compressing Agent Memory Without Losing the Original</a></h3>+ <p class="dek">How and why to compress an agent's long-term memory: the toolbox, the lossy trap, and keeping a lossless original beside a compact copy.</p>+ <div class="foot">+ <span class="tag kind">Internals</span>+ <a class="more" href="/blog/compressing-agent-memory/" aria-label="Read: Compressing Agent Memory Without Losing the Original">Read →</a>+ </div>+ </div>+</article><article class="card"><a class="thumb t-guides" href="/blog/the-multi-agent-cost-playbook/" aria-hidden="true" tabindex="-1"><span>GUIDES</span>diff --git a/docs/blog/the-hook-shim-pattern/index.html b/docs/blog/the-hook-shim-pattern/index.htmlindex 5f8cd3c6..a22d1089 100644--- a/docs/blog/the-hook-shim-pattern/index.html+++ b/docs/blog/the-hook-shim-pattern/index.html@@ -280,43 +280,43 @@ at the mercy of the bridge.<h2>Related in Internals</h2><div class="post-grid"><article class="card">- <a class="thumb t-internals" href="/blog/architecture-two-planes-one-renderer/" aria-hidden="true" tabindex="-1">+ <a class="thumb t-internals" href="/blog/compressing-agent-memory/" aria-hidden="true" tabindex="-1"><span>INTERNALS</span></a><div class="body"><div class="meta">- <time datetime="2026-06-04">Jun 4, 2026</time>+ <time datetime="2026-06-05">Jun 5, 2026</time><span class="dot" aria-hidden="true"></span><span>5 min</span></div>- <h3><a href="/blog/architecture-two-planes-one-renderer/">Munder Difflin's Architecture: Two Data Planes, One Renderer</a></h3>- <p class="dek">A walkthrough of the multi-agent harness architecture: a node-pty terminal plane and a hooks/hive event plane feeding one React + Pixi.js renderer.</p>+ <h3><a href="/blog/compressing-agent-memory/">Compressing Agent Memory Without Losing the Original</a></h3>+ <p class="dek">How and why to compress an agent's long-term memory: the toolbox, the lossy trap, and keeping a lossless original beside a compact copy.</p><div class="foot"><span class="tag kind">Internals</span>- <a class="more" href="/blog/architecture-two-planes-one-renderer/" aria-label="Read: Munder Difflin's Architecture: Two Data Planes, One Renderer">Read →</a>+ <a class="more" href="/blog/compressing-agent-memory/" aria-label="Read: Compressing Agent Memory Without Losing the Original">Read →</a></div></div></article><article class="card">- <a class="thumb t-internals" href="/blog/file-based-coordination-vs-message-queues/" aria-hidden="true" tabindex="-1">+ <a class="thumb t-internals" href="/blog/architecture-two-planes-one-renderer/" aria-hidden="true" tabindex="-1"><span>INTERNALS</span></a><div class="body"><div class="meta"><time datetime="2026-06-04">Jun 4, 2026</time><span class="dot" aria-hidden="true"></span>- <span>4 min</span>+ <span>5 min</span></div>- <h3><a href="/blog/file-based-coordination-vs-message-queues/">File-Based Coordination vs Message Queues for AI Agents</a></h3>- <p class="dek">Why a local agent hive coordinates through plain files, not Redis or RabbitMQ: the zero-ops wins, the real tradeoffs, and where files hit a ceiling.</p>+ <h3><a href="/blog/architecture-two-planes-one-renderer/">Munder Difflin's Architecture: Two Data Planes, One Renderer</a></h3>+ <p class="dek">A walkthrough of the multi-agent harness architecture: a node-pty terminal plane and a hooks/hive event plane feeding one React + Pixi.js renderer.</p><div class="foot"><span class="tag kind">Internals</span>- <a class="more" href="/blog/file-based-coordination-vs-message-queues/" aria-label="Read: File-Based Coordination vs Message Queues for AI Agents">Read →</a>+ <a class="more" href="/blog/architecture-two-planes-one-renderer/" aria-label="Read: Munder Difflin's Architecture: Two Data Planes, One Renderer">Read →</a></div></div></article><article class="card">- <a class="thumb t-internals" href="/blog/observability-for-agent-fleets/" aria-hidden="true" tabindex="-1">+ <a class="thumb t-internals" href="/blog/file-based-coordination-vs-message-queues/" aria-hidden="true" tabindex="-1"><span>INTERNALS</span></a><div class="body">@@ -325,11 +325,11 @@ at the mercy of the bridge.<span class="dot" aria-hidden="true"></span><span>4 min</span></div>- <h3><a href="/blog/observability-for-agent-fleets/">Observability for Agent Fleets: Seeing What Your Agents Do</a></h3>- <p class="dek">Agent fleet observability: the four questions your dashboard must answer about who's working, what they're doing, what it costs, and what they know.</p>+ <h3><a href="/blog/file-based-coordination-vs-message-queues/">File-Based Coordination vs Message Queues for AI Agents</a></h3>+ <p class="dek">Why a local agent hive coordinates through plain files, not Redis or RabbitMQ: the zero-ops wins, the real tradeoffs, and where files hit a ceiling.</p><div class="foot"><span class="tag kind">Internals</span>- <a class="more" href="/blog/observability-for-agent-fleets/" aria-label="Read: Observability for Agent Fleets: Seeing What Your Agents Do">Read →</a>+ <a class="more" href="/blog/file-based-coordination-vs-message-queues/" aria-label="Read: File-Based Coordination vs Message Queues for AI Agents">Read →</a></div></div></article>diff --git a/docs/blog/topics/comparisons/index.html b/docs/blog/topics/comparisons/index.htmlindex d67ec030..efb425e5 100644--- a/docs/blog/topics/comparisons/index.html+++ b/docs/blog/topics/comparisons/index.html@@ -70,7 +70,7 @@<a class="chip" href="/blog/topics/guides/">guides<span class="ct">21</span></a>- <a class="chip" href="/blog/topics/internals/">internals<span class="ct">14</span></a>+ <a class="chip" href="/blog/topics/internals/">internals<span class="ct">15</span></a><a class="chip" href="/blog/topics/concepts/">concepts<span class="ct">13</span></a>diff --git a/docs/blog/topics/concepts/index.html b/docs/blog/topics/concepts/index.htmlindex 9df89a01..3c0dc95b 100644--- a/docs/blog/topics/concepts/index.html+++ b/docs/blog/topics/concepts/index.html@@ -70,7 +70,7 @@<a class="chip" href="/blog/topics/guides/">guides<span class="ct">21</span></a>- <a class="chip" href="/blog/topics/internals/">internals<span class="ct">14</span></a>+ <a class="chip" href="/blog/topics/internals/">internals<span class="ct">15</span></a><a class="chip" href="/blog/topics/concepts/" aria-current="page">concepts<span class="ct">13</span></a>diff --git a/docs/blog/topics/guides/index.html b/docs/blog/topics/guides/index.htmlindex 9b1a7073..afbdf480 100644--- a/docs/blog/topics/guides/index.html+++ b/docs/blog/topics/guides/index.html@@ -70,7 +70,7 @@<a class="chip" href="/blog/topics/guides/" aria-current="page">guides<span class="ct">21</span></a>- <a class="chip" href="/blog/topics/internals/">internals<span class="ct">14</span></a>+ <a class="chip" href="/blog/topics/internals/">internals<span class="ct">15</span></a><a class="chip" href="/blog/topics/concepts/">concepts<span class="ct">13</span></a>diff --git a/docs/blog/topics/index.html b/docs/blog/topics/index.htmlindex 004658e2..95b70326 100644--- a/docs/blog/topics/index.html+++ b/docs/blog/topics/index.html@@ -79,7 +79,7 @@<span class="thumb t-internals" aria-hidden="true">INTERNALS</span><div class="body"><h3 style="color:var(--ink)">internals</h3>- <p class="dek">14 posts.</p>+ <p class="dek">15 posts.</p><div class="foot"><span class="more">View topic →</span></div></div></a>@@ -144,11 +144,11 @@<div class="filters" style="padding-top:36px"><span class="eyebrow" style="align-self:center">Tags</span>- <a class="chip" href="/blog/tags/multi-agent/">Multi-Agent<span class="ct">52</span></a>+ <a class="chip" href="/blog/tags/multi-agent/">Multi-Agent<span class="ct">53</span></a><a class="chip" href="/blog/tags/claude-code/">Claude Code<span class="ct">36</span></a>- <a class="chip" href="/blog/tags/internals/">Internals<span class="ct">20</span></a>+ <a class="chip" href="/blog/tags/internals/">Internals<span class="ct">21</span></a><a class="chip" href="/blog/tags/orchestration/">Orchestration<span class="ct">12</span></a>@@ -158,16 +158,16 @@<a class="chip" href="/blog/tags/concepts/">Concepts<span class="ct">10</span></a>+ <a class="chip" href="/blog/tags/memory/">Memory<span class="ct">8</span></a>+<a class="chip" href="/blog/tags/tools/">Tools<span class="ct">8</span></a><a class="chip" href="/blog/tags/getting-started/">Getting Started<span class="ct">7</span></a>- <a class="chip" href="/blog/tags/memory/">Memory<span class="ct">7</span></a>+ <a class="chip" href="/blog/tags/mempalace/">MemPalace<span class="ct">6</span></a><a class="chip" href="/blog/tags/automation/">Automation<span class="ct">6</span></a>- <a class="chip" href="/blog/tags/mempalace/">MemPalace<span class="ct">5</span></a>-<a class="chip" href="/blog/tags/local-first/">Local-First<span class="ct">5</span></a><a class="chip" href="/blog/tags/security/">Security<span class="ct">5</span></a>diff --git a/docs/blog/topics/internals/index.html b/docs/blog/topics/internals/index.htmlindex 14436731..2b8cb4b9 100644--- a/docs/blog/topics/internals/index.html+++ b/docs/blog/topics/internals/index.html@@ -63,14 +63,14 @@</nav><p class="eyebrow">Topic</p><h1>internals</h1>- <p class="lead">14 posts in this cluster.</p>+ <p class="lead">15 posts in this cluster.</p><div class="filters" role="navigation" aria-label="Topics"><a class="chip" href="/blog/">All</a><a class="chip" href="/blog/topics/guides/">guides<span class="ct">21</span></a>- <a class="chip" href="/blog/topics/internals/" aria-current="page">internals<span class="ct">14</span></a>+ <a class="chip" href="/blog/topics/internals/" aria-current="page">internals<span class="ct">15</span></a><a class="chip" href="/blog/topics/concepts/">concepts<span class="ct">13</span></a>@@ -90,6 +90,24 @@<section class="wrap"><div class="post-grid"><article class="card">+ <a class="thumb t-internals" href="/blog/compressing-agent-memory/" aria-hidden="true" tabindex="-1">+ <span>INTERNALS</span>+ </a>+ <div class="body">+ <div class="meta">+ <time datetime="2026-06-05">Jun 5, 2026</time>+ <span class="dot" aria-hidden="true"></span>+ <span>5 min</span>+ </div>+ <h3><a href="/blog/compressing-agent-memory/">Compressing Agent Memory Without Losing the Original</a></h3>+ <p class="dek">How and why to compress an agent's long-term memory: the toolbox, the lossy trap, and keeping a lossless original beside a compact copy.</p>+ <div class="foot">+ <span class="tag kind">Internals</span>+ <a class="more" href="/blog/compressing-agent-memory/" aria-label="Read: Compressing Agent Memory Without Losing the Original">Read →</a>+ </div>+ </div>+</article>+<article class="card"><a class="thumb t-internals" href="/blog/architecture-two-planes-one-renderer/" aria-hidden="true" tabindex="-1"><span>INTERNALS</span></a>diff --git a/docs/blog/topics/memory/index.html b/docs/blog/topics/memory/index.htmlindex 4e23bcd8..dc257de7 100644--- a/docs/blog/topics/memory/index.html+++ b/docs/blog/topics/memory/index.html@@ -70,7 +70,7 @@<a class="chip" href="/blog/topics/guides/">guides<span class="ct">21</span></a>- <a class="chip" href="/blog/topics/internals/">internals<span class="ct">14</span></a>+ <a class="chip" href="/blog/topics/internals/">internals<span class="ct">15</span></a><a class="chip" href="/blog/topics/concepts/">concepts<span class="ct">13</span></a>diff --git a/docs/blog/topics/orchestration/index.html b/docs/blog/topics/orchestration/index.htmlindex 221090e5..cbefb7b2 100644--- a/docs/blog/topics/orchestration/index.html+++ b/docs/blog/topics/orchestration/index.html@@ -70,7 +70,7 @@<a class="chip" href="/blog/topics/guides/">guides<span class="ct">21</span></a>- <a class="chip" href="/blog/topics/internals/">internals<span class="ct">14</span></a>+ <a class="chip" href="/blog/topics/internals/">internals<span class="ct">15</span></a><a class="chip" href="/blog/topics/concepts/">concepts<span class="ct">13</span></a>diff --git a/docs/blog/topics/story/index.html b/docs/blog/topics/story/index.htmlindex e03fa66b..9084bd13 100644--- a/docs/blog/topics/story/index.html+++ b/docs/blog/topics/story/index.html@@ -70,7 +70,7 @@<a class="chip" href="/blog/topics/guides/">guides<span class="ct">21</span></a>- <a class="chip" href="/blog/topics/internals/">internals<span class="ct">14</span></a>+ <a class="chip" href="/blog/topics/internals/">internals<span class="ct">15</span></a><a class="chip" href="/blog/topics/concepts/">concepts<span class="ct">13</span></a>diff --git a/docs/blog/topics/use-cases/index.html b/docs/blog/topics/use-cases/index.htmlindex 6444fdc4..f88f5068 100644--- a/docs/blog/topics/use-cases/index.html+++ b/docs/blog/topics/use-cases/index.html@@ -70,7 +70,7 @@<a class="chip" href="/blog/topics/guides/">guides<span class="ct">21</span></a>- <a class="chip" href="/blog/topics/internals/">internals<span class="ct">14</span></a>+ <a class="chip" href="/blog/topics/internals/">internals<span class="ct">15</span></a><a class="chip" href="/blog/topics/concepts/">concepts<span class="ct">13</span></a>diff --git a/docs/blog/visualizing-ai-agents-pixijs/index.html b/docs/blog/visualizing-ai-agents-pixijs/index.htmlindex 7c076b08..13d9ebad 100644--- a/docs/blog/visualizing-ai-agents-pixijs/index.html+++ b/docs/blog/visualizing-ai-agents-pixijs/index.html@@ -281,43 +281,43 @@ to see your agents at work; it’s free and open source.</p><h2>Related in Internals</h2><div class="post-grid"><article class="card">- <a class="thumb t-internals" href="/blog/architecture-two-planes-one-renderer/" aria-hidden="true" tabindex="-1">+ <a class="thumb t-internals" href="/blog/compressing-agent-memory/" aria-hidden="true" tabindex="-1"><span>INTERNALS</span></a><div class="body"><div class="meta">- <time datetime="2026-06-04">Jun 4, 2026</time>+ <time datetime="2026-06-05">Jun 5, 2026</time><span class="dot" aria-hidden="true"></span><span>5 min</span></div>- <h3><a href="/blog/architecture-two-planes-one-renderer/">Munder Difflin's Architecture: Two Data Planes, One Renderer</a></h3>- <p class="dek">A walkthrough of the multi-agent harness architecture: a node-pty terminal plane and a hooks/hive event plane feeding one React + Pixi.js renderer.</p>+ <h3><a href="/blog/compressing-agent-memory/">Compressing Agent Memory Without Losing the Original</a></h3>+ <p class="dek">How and why to compress an agent's long-term memory: the toolbox, the lossy trap, and keeping a lossless original beside a compact copy.</p><div class="foot"><span class="tag kind">Internals</span>- <a class="more" href="/blog/architecture-two-planes-one-renderer/" aria-label="Read: Munder Difflin's Architecture: Two Data Planes, One Renderer">Read →</a>+ <a class="more" href="/blog/compressing-agent-memory/" aria-label="Read: Compressing Agent Memory Without Losing the Original">Read →</a></div></div></article><article class="card">- <a class="thumb t-internals" href="/blog/file-based-coordination-vs-message-queues/" aria-hidden="true" tabindex="-1">+ <a class="thumb t-internals" href="/blog/architecture-two-planes-one-renderer/" aria-hidden="true" tabindex="-1"><span>INTERNALS</span></a><div class="body"><div class="meta"><time datetime="2026-06-04">Jun 4, 2026</time><span class="dot" aria-hidden="true"></span>- <span>4 min</span>+ <span>5 min</span></div>- <h3><a href="/blog/file-based-coordination-vs-message-queues/">File-Based Coordination vs Message Queues for AI Agents</a></h3>- <p class="dek">Why a local agent hive coordinates through plain files, not Redis or RabbitMQ: the zero-ops wins, the real tradeoffs, and where files hit a ceiling.</p>+ <h3><a href="/blog/architecture-two-planes-one-renderer/">Munder Difflin's Architecture: Two Data Planes, One Renderer</a></h3>+ <p class="dek">A walkthrough of the multi-agent harness architecture: a node-pty terminal plane and a hooks/hive event plane feeding one React + Pixi.js renderer.</p><div class="foot"><span class="tag kind">Internals</span>- <a class="more" href="/blog/file-based-coordination-vs-message-queues/" aria-label="Read: File-Based Coordination vs Message Queues for AI Agents">Read →</a>+ <a class="more" href="/blog/architecture-two-planes-one-renderer/" aria-label="Read: Munder Difflin's Architecture: Two Data Planes, One Renderer">Read →</a></div></div></article><article class="card">- <a class="thumb t-internals" href="/blog/observability-for-agent-fleets/" aria-hidden="true" tabindex="-1">+ <a class="thumb t-internals" href="/blog/file-based-coordination-vs-message-queues/" aria-hidden="true" tabindex="-1"><span>INTERNALS</span></a><div class="body">@@ -326,11 +326,11 @@ to see your agents at work; it’s free and open source.</p><span class="dot" aria-hidden="true"></span><span>4 min</span></div>- <h3><a href="/blog/observability-for-agent-fleets/">Observability for Agent Fleets: Seeing What Your Agents Do</a></h3>- <p class="dek">Agent fleet observability: the four questions your dashboard must answer about who's working, what they're doing, what it costs, and what they know.</p>+ <h3><a href="/blog/file-based-coordination-vs-message-queues/">File-Based Coordination vs Message Queues for AI Agents</a></h3>+ <p class="dek">Why a local agent hive coordinates through plain files, not Redis or RabbitMQ: the zero-ops wins, the real tradeoffs, and where files hit a ceiling.</p><div class="foot"><span class="tag kind">Internals</span>- <a class="more" href="/blog/observability-for-agent-fleets/" aria-label="Read: Observability for Agent Fleets: Seeing What Your Agents Do">Read →</a>+ <a class="more" href="/blog/file-based-coordination-vs-message-queues/" aria-label="Read: File-Based Coordination vs Message Queues for AI Agents">Read →</a></div></div></article>diff --git a/docs/sitemap.xml b/docs/sitemap.xmlindex 9ab8949e..2a84991a 100644--- a/docs/sitemap.xml+++ b/docs/sitemap.xml@@ -23,6 +23,12 @@<changefreq>monthly</changefreq><priority>0.8</priority></url>+ <url>+ <loc>https://munderdiffl.in/blog/compressing-agent-memory/</loc>+ <lastmod>2026-06-05</lastmod>+ <changefreq>monthly</changefreq>+ <priority>0.8</priority>+ </url><url><loc>https://munderdiffl.in/blog/generative-engine-optimization-techniques/</loc><lastmod>2026-06-05</lastmod>@@ -494,11 +500,11 @@<url><loc>https://munderdiffl.in/blog/tags/guides/</loc><changefreq>weekly</changefreq><priority>0.3</priority></url><url><loc>https://munderdiffl.in/blog/tags/comparisons/</loc><changefreq>weekly</changefreq><priority>0.3</priority></url><url><loc>https://munderdiffl.in/blog/tags/concepts/</loc><changefreq>weekly</changefreq><priority>0.3</priority></url>+ <url><loc>https://munderdiffl.in/blog/tags/memory/</loc><changefreq>weekly</changefreq><priority>0.3</priority></url><url><loc>https://munderdiffl.in/blog/tags/tools/</loc><changefreq>weekly</changefreq><priority>0.3</priority></url><url><loc>https://munderdiffl.in/blog/tags/getting-started/</loc><changefreq>weekly</changefreq><priority>0.3</priority></url>- <url><loc>https://munderdiffl.in/blog/tags/memory/</loc><changefreq>weekly</changefreq><priority>0.3</priority></url>- <url><loc>https://munderdiffl.in/blog/tags/automation/</loc><changefreq>weekly</changefreq><priority>0.3</priority></url><url><loc>https://munderdiffl.in/blog/tags/mempalace/</loc><changefreq>weekly</changefreq><priority>0.3</priority></url>+ <url><loc>https://munderdiffl.in/blog/tags/automation/</loc><changefreq>weekly</changefreq><priority>0.3</priority></url><url><loc>https://munderdiffl.in/blog/tags/local-first/</loc><changefreq>weekly</changefreq><priority>0.3</priority></url><url><loc>https://munderdiffl.in/blog/tags/security/</loc><changefreq>weekly</changefreq><priority>0.3</priority></url><url><loc>https://munderdiffl.in/blog/tags/autonomous/</loc><changefreq>weekly</changefreq><priority>0.3</priority></url>