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: HITL post — human-in-the-loop-approving-ai-agents (56 → 57)
Chaitanya Giri4425f195544a
diff --git a/blog/src/posts/human-in-the-loop-approving-ai-agents.md b/blog/src/posts/human-in-the-loop-approving-ai-agents.mdnew file mode 100644index 00000000..ca8fd1af--- /dev/null+++ b/blog/src/posts/human-in-the-loop-approving-ai-agents.md@@ -0,0 +1,134 @@+---+title: "Approving AI Agents: Human-in-the-Loop Without a Queue"+description: "How to approve what AI agents do without a parallel approval queue: native permission prompts, routing 'to: human' to the orchestrator, remote approval."+date: 2026-06-04+category: guides+categoryLabel: Guides+type: Technical+primaryKeyword: "approving ai agents"+secondaryKeywords: ["human in the loop ai agents", "ai agent approvals", "approve ai agent actions", "human in the loop approval"]+tags: ["Human-in-the-Loop", "Hive", "Orchestration", "Internals"]+author:+ name: Chaitanya Giri+ initials: CG+faq:+ - q: "How do you put a human in the loop on AI agent actions?"+ a: "Approve where the work already happens. Modern coding agents (Claude Code among them) already pause and ask permission before risky actions, right in their own session. The reliable pattern is to lean on that native prompt rather than building a separate approval queue — and to route any explicit 'I need a human' decision to a single person or their proxy, so there's one place an answer comes from."+ - q: "Why not build a dedicated approval queue for agents?"+ a: "Because it duplicates state. A custom queue becomes a second source of truth that can drift from the agent's actual state and adds its own failure modes — Munder Difflin's old in-app queue could even re-queue an item when you approved it. Removing the queue in favor of native permission prompts deleted the bug along with the abstraction."+ - q: "Can you approve an AI agent's action remotely?"+ a: "Yes. Because approval rides on the native permission prompt in the agent's session, it can be approved wherever you can reach that session — including from your phone via /remote-control. A blocking action waits for your yes; you don't have to be at your desk to give it."+---++<div class="callout tldr"><span class="ic">TL;DR</span><p>The right way to put a human in the loop on+agent actions isn't a <strong>custom approval queue</strong> bolted onto your system — it's to+<strong>approve where the work already happens</strong>. Munder Difflin <em>removed</em> its in-app+approval queue in v0.1.7 in favor of <strong>native Claude Code permission prompts</strong>: the agent+pauses in its own terminal, you approve (even from your phone via <code>/remote-control</code>), and any+explicit "I need a human" message routes to the <strong>orchestrator — the human's proxy on the+floor</strong>. No second source of truth, no queue to drift, one less bug.</p></div>++Once you let AI agents <em>act</em> — run commands, edit files, touch infrastructure — you need a way to+keep a human in the loop on the decisions that matter. The obvious design is a queue: agents drop+"please approve" items into a panel, a human clicks yes or no. It's also, in our experience, the wrong+abstraction. This post is about a better one, grounded in how a real+[multi-agent harness](/#what) does it.++## The approval problem++An autonomous agent is valuable precisely because it doesn't stop to ask you about every line. But some+actions deserve a human: deleting data, spending real money, changing scope, force-pushing, anything+irreversible. So you want a narrow gate — most actions flow, a few pause for a person — without turning+the human into a bottleneck who rubber-stamps a hundred prompts a day.++The question is *where* that gate lives. And the tempting answer — a bespoke approval queue — quietly+creates more problems than it solves.++## The trap: a parallel approval queue++A custom approval queue is a second copy of state. The agent has a real, live situation in its session;+the queue has a *representation* of that situation in a panel. Now you own the hard part of distributed+systems: keeping two sources of truth in sync. They drift. The queue shows an item the agent already+moved past; an approval lands but the agent's context has changed; the panel and the session disagree+about what "yes" even means.++It also adds its own failure modes. Munder Difflin shipped exactly this at first — a floating approvals+panel — and learned the lesson the hard way: approving an item could **re-queue it**, because resolving+an approval re-routed the message back into the queue. The+[v0.1.7 release](https://github.com/chaitanyagiri/munder-difflin/blob/main/CHANGELOG.md) notes it plainly:+"Moving to native HITL removes the panel and the bug." Deleting the abstraction deleted the class of bug+with it.++## The native approach: approve where the work happens++Here's the shift. Coding agents *already have* a human-in-the-loop mechanism — the **permission prompt**.+Claude Code pauses before a sensitive tool call and asks, right there in the agent's own terminal. That+prompt is the real gate: it blocks the actual action, in the actual session, with the actual context.+There's nothing to keep in sync because there's only one place the decision lives.++So the better design is to *lean on the native prompt* instead of reinventing it. As the harness's hive+module puts it in its own header comment:++> Human-in-the-loop is native to each agent's Claude Code session: permission prompts surface in the+> agent's own terminal … The hive keeps no separate approval queue.++That's the whole philosophy in two sentences. The agent stops itself at the dangerous step; a human+answers in the place the agent stopped. No panel, no mirror of state, no re-queue bug.++## Routing a "to: human" decision++There's a second kind of human-in-the-loop, though: not "approve this tool call" but "I'm an agent and I+genuinely need a *person* to decide something" — an ambiguity, a conflict, a scope call. Where does that+go when there's no human queue?++To the **orchestrator**. In a hive, the [god/orchestrator](/#how) (we call him Michael) is the human's+proxy on the floor — so a message addressed to `"human"` is simply routed there. (Triage, delegation,+and escalation are the orchestrator's whole job —+[how the god orchestrator works](/blog/how-the-god-orchestrator-works/).) The router does it with one+line:++```ts+// src/main/hive.ts — resolve the recipient+const resolveTo = (to) => (to === 'human' || to === 'god' ? godId : to);+```++Both `"god"` and `"human"` collapse to the orchestrator's id. The orchestrator triages it, answers what+it can, and escalates to the actual person — natively, in his own session — only for the genuinely+critical calls. A `needsHuman` flag still rides along, but it's **cosmetic**: it tints the message+envelope on the office-floor visualization. As the code comments note, it's "Cosmetic only — no queue+behind it." The system even guards against a `god → "human"` message looping back to itself. One+recipient, one source of truth, no parallel inbox to babysit.++## Approve from anywhere++A nice property falls out of doing it this way: because approval rides on the native session prompt,+**you can approve from wherever you can reach the session.** Munder Difflin's `/remote-control` lets the+human approve a blocking action from their phone — the agent waits, you tap yes on the train, it+continues. A queue would have forced you to build remote approval as yet another feature; native HITL+gets it almost for free, because you're just answering the prompt the agent is already blocked on.++## When to actually ask a human++The mechanism only helps if the *policy* is right. The aim is a team that stays unblocked, with the+human reserved for decisions that truly need them:++- **Destructive or irreversible actions** — dropping data, deleting branches, force-pushing.+- **Spending real money** — provisioning paid resources, large API spend.+- **Scope changes** — anything that redefines what "done" means.+- **Unresolvable conflicts** — two agents at an impasse the orchestrator can't settle.++Everything else should flow. Over-gating trains the human to click "approve" without reading, which is+worse than no gate at all. Pair a tight policy with an+[append-only audit log](/blog/append-only-event-log-agents/) and every approval has a precise,+timestamped answer to "who said yes to that?" — without a queue in sight. (For the broader case for+keeping people in the loop, see [human-in-the-loop AI agents](/blog/human-in-the-loop-ai-agents/).)++## The takeaway++Human-in-the-loop is a *placement* problem more than a UI problem. Put the gate where the action already+blocks — the native permission prompt — and route genuine "I need a person" decisions to a single proxy,+the orchestrator. You get approvals that can't drift, that work remotely, and that don't carry their own+bugs. The best approval queue is the one you didn't build.++Want to see native, no-queue approvals — and a hive that routes its own escalations to an orchestrator+you talk to? You can [download Munder Difflin](/#install) free; it's open source.diff --git a/docs/blog/feed.xml b/docs/blog/feed.xmlindex ca983331..cb2fc85e 100644--- a/docs/blog/feed.xml+++ b/docs/blog/feed.xml@@ -111,6 +111,14 @@<updated>2026-06-04T00:00:00.000Z</updated><category term="Concepts" /><summary>The discipline that makes an autonomous coding agent trustworthy: prove every claim, reproduce the green, and check the fix — not just the intent.</summary></entry>+ <entry>+ <title>Approving AI Agents: Human-in-the-Loop Without a Queue</title>+ <link href="https://munderdiffl.in/blog/human-in-the-loop-approving-ai-agents/" />+ <id>https://munderdiffl.in/blog/human-in-the-loop-approving-ai-agents/</id>+ <published>2026-06-04T00:00:00.000Z</published>+ <updated>2026-06-04T00:00:00.000Z</updated><category term="Guides" />+ <summary>How to approve what AI agents do without a parallel approval queue: native permission prompts, routing 'to: human' to the orchestrator, remote approval.</summary>+ </entry><entry><title>Keep Your Agent's Semantic Memory Clean: Don't Index the Config</title><link href="https://munderdiffl.in/blog/keep-agent-semantic-memory-clean/" />diff --git a/docs/blog/how-ai-agents-verify-their-own-work/index.html b/docs/blog/how-ai-agents-verify-their-own-work/index.htmlindex b5d7c618..9eb7cdea 100644--- a/docs/blog/how-ai-agents-verify-their-own-work/index.html+++ b/docs/blog/how-ai-agents-verify-their-own-work/index.html@@ -267,7 +267,7 @@ and open source.</p></div><footer class="post-foot wrap"><div class="prevnext">- <a class="pn prev" href="/blog/keep-agent-semantic-memory-clean/"><span class="k">← Older</span><span class="t">Keep Your Agent's Semantic Memory Clean: Don't Index the Config</span></a>+ <a class="pn prev" href="/blog/human-in-the-loop-approving-ai-agents/"><span class="k">← Older</span><span class="t">Approving AI Agents: Human-in-the-Loop Without a Queue</span></a><a class="pn next" href="/blog/how-agents-remember-semantic-memory/"><span class="k">Newer →</span><span class="t">How AI Agents Remember: Semantic Memory in a Hive</span></a></div><section class="related">diff --git a/docs/blog/human-in-the-loop-approving-ai-agents/index.html b/docs/blog/human-in-the-loop-approving-ai-agents/index.htmlnew file mode 100644index 00000000..f728da81--- /dev/null+++ b/docs/blog/human-in-the-loop-approving-ai-agents/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>Approving AI Agents: Human-in-the-Loop Without a Queue — Munder Difflin Blog</title>+<meta name="description" content="How to approve what AI agents do without a parallel approval queue: native permission prompts, routing 'to: human' to the orchestrator, remote approval." />+<link rel="canonical" href="https://munderdiffl.in/blog/human-in-the-loop-approving-ai-agents/" />++<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="Approving AI Agents: Human-in-the-Loop Without a Queue — Munder Difflin Blog" />+<meta property="og:description" content="How to approve what AI agents do without a parallel approval queue: native permission prompts, routing 'to: human' to the orchestrator, remote approval." />+<meta property="og:url" content="https://munderdiffl.in/blog/human-in-the-loop-approving-ai-agents/" />+<meta property="og:image" content="https://munderdiffl.in/media/og.png" />+<meta property="og:locale" content="en_US" />+<meta property="article:published_time" content="2026-06-04T00:00:00.000Z" />+<meta property="article:author" content="Chaitanya Giri" />+<meta property="article:tag" content="Human-in-the-Loop" />+<meta property="article:tag" content="Hive" />+<meta property="article:tag" content="Orchestration" />+<meta property="article:tag" content="Internals" />+++<!-- Twitter -->+<meta name="twitter:card" content="summary_large_image" />+<meta name="twitter:title" content="Approving AI Agents: Human-in-the-Loop Without a Queue — Munder Difflin Blog" />+<meta name="twitter:description" content="How to approve what AI agents do without a parallel approval queue: native permission prompts, routing 'to: human' to the orchestrator, remote approval." />+<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": "Approving AI Agents: Human-in-the-Loop Without a Queue",+ "description": "How to approve what AI agents do without a parallel approval queue: native permission prompts, routing 'to: human' to the orchestrator, remote approval.",+ "image": ["https://munderdiffl.in/media/og.png"],+ "datePublished": "2026-06-04T00:00:00.000Z",+ "dateModified": "2026-06-04T00:00:00.000Z",+ "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/human-in-the-loop-approving-ai-agents/" },+ "keywords": "approving ai agents, human in the loop ai agents, ai agent approvals, approve ai agent actions, human in the loop approval",+ "articleSection": "Guides",+ "inLanguage": "en-US",+ "url": "https://munderdiffl.in/blog/human-in-the-loop-approving-ai-agents/"+}+</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": "Approving AI Agents: Human-in-the-Loop Without a Queue", "item": "https://munderdiffl.in/blog/human-in-the-loop-approving-ai-agents/" }+ ]+}+</script>++<script type="application/ld+json">+{+ "@context": "https://schema.org",+ "@type": "FAQPage",+ "mainEntity": [+ { "@type": "Question", "name": "How do you put a human in the loop on AI agent actions?", "acceptedAnswer": { "@type": "Answer", "text": "Approve where the work already happens. Modern coding agents (Claude Code among them) already pause and ask permission before risky actions, right in their own session. The reliable pattern is to lean on that native prompt rather than building a separate approval queue — and to route any explicit 'I need a human' decision to a single person or their proxy, so there's one place an answer comes from." } },+ { "@type": "Question", "name": "Why not build a dedicated approval queue for agents?", "acceptedAnswer": { "@type": "Answer", "text": "Because it duplicates state. A custom queue becomes a second source of truth that can drift from the agent's actual state and adds its own failure modes — Munder Difflin's old in-app queue could even re-queue an item when you approved it. Removing the queue in favor of native permission prompts deleted the bug along with the abstraction." } },+ { "@type": "Question", "name": "Can you approve an AI agent's action remotely?", "acceptedAnswer": { "@type": "Answer", "text": "Yes. Because approval rides on the native permission prompt in the agent's session, it can be approved wherever you can reach that session — including from your phone via /remote-control. A blocking action waits for your yes; you don't have to be at your desk to give it." } }++ ]+}+</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/guides/">Guides</a>+ </nav>+ <h1>Approving AI Agents: Human-in-the-Loop Without a Queue</h1>+ <p class="dek">How to approve what AI agents do without a parallel approval queue: native permission prompts, routing 'to: human' to the orchestrator, remote approval.</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="2026-06-04">Jun 4, 2026</time>+ <span class="dot" aria-hidden="true">·</span>+ <span>5 min read</span>+ <span class="tag kind">Guides</span>+ </div>+ </header>++ <div class="post-wrap wrap">+ <div class="prose">+ <div class="callout tldr"><span class="ic">TL;DR</span><p>The right way to put a human in the loop on+agent actions isn't a <strong>custom approval queue</strong> bolted onto your system — it's to+<strong>approve where the work already happens</strong>. Munder Difflin <em>removed</em> its in-app+approval queue in v0.1.7 in favor of <strong>native Claude Code permission prompts</strong>: the agent+pauses in its own terminal, you approve (even from your phone via <code>/remote-control</code>), and any+explicit "I need a human" message routes to the <strong>orchestrator — the human's proxy on the+floor</strong>. No second source of truth, no queue to drift, one less bug.</p></div>+<p>Once you let AI agents <em>act</em> — run commands, edit files, touch infrastructure — you need a way to+keep a human in the loop on the decisions that matter. The obvious design is a queue: agents drop+“please approve” items into a panel, a human clicks yes or no. It’s also, in our experience, the wrong+abstraction. This post is about a better one, grounded in how a real+<a href="/#what">multi-agent harness</a> does it.</p>+<h2 id="the-approval-problem" tabindex="-1">The approval problem <a class="anchor" href="#the-approval-problem" aria-hidden="true">#</a></h2>+<p>An autonomous agent is valuable precisely because it doesn’t stop to ask you about every line. But some+actions deserve a human: deleting data, spending real money, changing scope, force-pushing, anything+irreversible. So you want a narrow gate — most actions flow, a few pause for a person — without turning+the human into a bottleneck who rubber-stamps a hundred prompts a day.</p>+<p>The question is <em>where</em> that gate lives. And the tempting answer — a bespoke approval queue — quietly+creates more problems than it solves.</p>+<h2 id="the-trap-a-parallel-approval-queue" tabindex="-1">The trap: a parallel approval queue <a class="anchor" href="#the-trap-a-parallel-approval-queue" aria-hidden="true">#</a></h2>+<p>A custom approval queue is a second copy of state. The agent has a real, live situation in its session;+the queue has a <em>representation</em> of that situation in a panel. Now you own the hard part of distributed+systems: keeping two sources of truth in sync. They drift. The queue shows an item the agent already+moved past; an approval lands but the agent’s context has changed; the panel and the session disagree+about what “yes” even means.</p>+<p>It also adds its own failure modes. Munder Difflin shipped exactly this at first — a floating approvals+panel — and learned the lesson the hard way: approving an item could <strong>re-queue it</strong>, because resolving+an approval re-routed the message back into the queue. The+<a href="https://github.com/chaitanyagiri/munder-difflin/blob/main/CHANGELOG.md">v0.1.7 release</a> notes it plainly:+“Moving to native HITL removes the panel and the bug.” Deleting the abstraction deleted the class of bug+with it.</p>+<h2 id="the-native-approach-approve-where-the-work-happens" tabindex="-1">The native approach: approve where the work happens <a class="anchor" href="#the-native-approach-approve-where-the-work-happens" aria-hidden="true">#</a></h2>+<p>Here’s the shift. Coding agents <em>already have</em> a human-in-the-loop mechanism — the <strong>permission prompt</strong>.+Claude Code pauses before a sensitive tool call and asks, right there in the agent’s own terminal. That+prompt is the real gate: it blocks the actual action, in the actual session, with the actual context.+There’s nothing to keep in sync because there’s only one place the decision lives.</p>+<p>So the better design is to <em>lean on the native prompt</em> instead of reinventing it. As the harness’s hive+module puts it in its own header comment:</p>+<blockquote>+<p>Human-in-the-loop is native to each agent’s Claude Code session: permission prompts surface in the+agent’s own terminal … The hive keeps no separate approval queue.</p>+</blockquote>+<p>That’s the whole philosophy in two sentences. The agent stops itself at the dangerous step; a human+answers in the place the agent stopped. No panel, no mirror of state, no re-queue bug.</p>+<h2 id="routing-a-to-human-decision" tabindex="-1">Routing a “to: human” decision <a class="anchor" href="#routing-a-to-human-decision" aria-hidden="true">#</a></h2>+<p>There’s a second kind of human-in-the-loop, though: not “approve this tool call” but “I’m an agent and I+genuinely need a <em>person</em> to decide something” — an ambiguity, a conflict, a scope call. Where does that+go when there’s no human queue?</p>+<p>To the <strong>orchestrator</strong>. In a hive, the <a href="/#how">god/orchestrator</a> (we call him Michael) is the human’s+proxy on the floor — so a message addressed to <code>"human"</code> is simply routed there. (Triage, delegation,+and escalation are the orchestrator’s whole job —+<a href="/blog/how-the-god-orchestrator-works/">how the god orchestrator works</a>.) The router does it with one+line:</p>+<pre class="language-ts"><code class="language-ts"><span class="token comment">// src/main/hive.ts — resolve the recipient</span>+<span class="token keyword">const</span> <span class="token function-variable function">resolveTo</span> <span class="token operator">=</span> <span class="token punctuation">(</span>to<span class="token punctuation">)</span> <span class="token operator">=></span> <span class="token punctuation">(</span>to <span class="token operator">===</span> <span class="token string">'human'</span> <span class="token operator">||</span> to <span class="token operator">===</span> <span class="token string">'god'</span> <span class="token operator">?</span> godId <span class="token operator">:</span> to<span class="token punctuation">)</span><span class="token punctuation">;</span></code></pre>+<p>Both <code>"god"</code> and <code>"human"</code> collapse to the orchestrator’s id. The orchestrator triages it, answers what+it can, and escalates to the actual person — natively, in his own session — only for the genuinely+critical calls. A <code>needsHuman</code> flag still rides along, but it’s <strong>cosmetic</strong>: it tints the message+envelope on the office-floor visualization. As the code comments note, it’s “Cosmetic only — no queue+behind it.” The system even guards against a <code>god → "human"</code> message looping back to itself. One+recipient, one source of truth, no parallel inbox to babysit.</p>+<h2 id="approve-from-anywhere" tabindex="-1">Approve from anywhere <a class="anchor" href="#approve-from-anywhere" aria-hidden="true">#</a></h2>+<p>A nice property falls out of doing it this way: because approval rides on the native session prompt,+<strong>you can approve from wherever you can reach the session.</strong> Munder Difflin’s <code>/remote-control</code> lets the+human approve a blocking action from their phone — the agent waits, you tap yes on the train, it+continues. A queue would have forced you to build remote approval as yet another feature; native HITL+gets it almost for free, because you’re just answering the prompt the agent is already blocked on.</p>+<h2 id="when-to-actually-ask-a-human" tabindex="-1">When to actually ask a human <a class="anchor" href="#when-to-actually-ask-a-human" aria-hidden="true">#</a></h2>+<p>The mechanism only helps if the <em>policy</em> is right. The aim is a team that stays unblocked, with the+human reserved for decisions that truly need them:</p>+<ul>+<li><strong>Destructive or irreversible actions</strong> — dropping data, deleting branches, force-pushing.</li>+<li><strong>Spending real money</strong> — provisioning paid resources, large API spend.</li>+<li><strong>Scope changes</strong> — anything that redefines what “done” means.</li>+<li><strong>Unresolvable conflicts</strong> — two agents at an impasse the orchestrator can’t settle.</li>+</ul>+<p>Everything else should flow. Over-gating trains the human to click “approve” without reading, which is+worse than no gate at all. Pair a tight policy with an+<a href="/blog/append-only-event-log-agents/">append-only audit log</a> and every approval has a precise,+timestamped answer to “who said yes to that?” — without a queue in sight. (For the broader case for+keeping people in the loop, see <a href="/blog/human-in-the-loop-ai-agents/">human-in-the-loop AI agents</a>.)</p>+<h2 id="the-takeaway" tabindex="-1">The takeaway <a class="anchor" href="#the-takeaway" aria-hidden="true">#</a></h2>+<p>Human-in-the-loop is a <em>placement</em> problem more than a UI problem. Put the gate where the action already+blocks — the native permission prompt — and route genuine “I need a person” decisions to a single proxy,+the orchestrator. You get approvals that can’t drift, that work remotely, and that don’t carry their own+bugs. The best approval queue is the one you didn’t build.</p>+<p>Want to see native, no-queue approvals — and a hive that routes its own escalations to an orchestrator+you talk to? You can <a href="/#install">download Munder Difflin</a> free; it’s open source.</p>++++ <section class="faq" aria-label="Frequently asked questions">+ <h2 id="faq">FAQ</h2>++ <div class="qa">+ <p class="q">How do you put a human in the loop on AI agent actions?</p>+ <p class="a">Approve where the work already happens. Modern coding agents (Claude Code among them) already pause and ask permission before risky actions, right in their own session. The reliable pattern is to lean on that native prompt rather than building a separate approval queue — and to route any explicit 'I need a human' decision to a single person or their proxy, so there's one place an answer comes from.</p>+ </div>++ <div class="qa">+ <p class="q">Why not build a dedicated approval queue for agents?</p>+ <p class="a">Because it duplicates state. A custom queue becomes a second source of truth that can drift from the agent's actual state and adds its own failure modes — Munder Difflin's old in-app queue could even re-queue an item when you approved it. Removing the queue in favor of native permission prompts deleted the bug along with the abstraction.</p>+ </div>++ <div class="qa">+ <p class="q">Can you approve an AI agent's action remotely?</p>+ <p class="a">Yes. Because approval rides on the native permission prompt in the agent's session, it can be approved wherever you can reach that session — including from your phone via /remote-control. A blocking action waits for your yes; you don't have to be at your desk to give it.</p>+ </div>++ </section>++++ <hr />+ <div class="share">+ <span class="lbl">Tags</span>+ <a class="tag" href="/blog/tags/human-in-the-loop/">Human-in-the-Loop</a><a class="tag" href="/blog/tags/hive/">Hive</a><a class="tag" href="/blog/tags/orchestration/">Orchestration</a><a class="tag" href="/blog/tags/internals/">Internals</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="#the-approval-problem">The approval problem</a></li><li class="lvl-2"><a href="#the-trap-a-parallel-approval-queue">The trap: a parallel approval queue</a></li><li class="lvl-2"><a href="#the-native-approach-approve-where-the-work-happens">The native approach: approve where the work happens</a></li><li class="lvl-2"><a href="#routing-a-to-human-decision">Routing a “to: human” decision</a></li><li class="lvl-2"><a href="#approve-from-anywhere">Approve from anywhere</a></li><li class="lvl-2"><a href="#when-to-actually-ask-a-human">When to actually ask a human</a></li><li class="lvl-2"><a href="#the-takeaway">The takeaway</a></li>+ </ol>+ </aside>++ </div>++ <footer class="post-foot wrap"><div class="prevnext">+ <a class="pn prev" href="/blog/keep-agent-semantic-memory-clean/"><span class="k">← Older</span><span class="t">Keep Your Agent's Semantic Memory Clean: Don't Index the Config</span></a>+ <a class="pn next" href="/blog/how-ai-agents-verify-their-own-work/"><span class="k">Newer →</span><span class="t">How AI Agents Verify Their Own Work (Before Saying 'Done')</span></a>+ </div>+ <section class="related">+ <h2>Related in Guides</h2>+ <div class="post-grid">+ <article class="card">+ <a class="thumb t-guides" href="/blog/how-to-install-and-use-munder-difflin/" aria-hidden="true" tabindex="-1">+ <span>GUIDES</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>9 min</span>+ </div>+ <h3><a href="/blog/how-to-install-and-use-munder-difflin/">How to Install and Use Munder Difflin</a></h3>+ <p class="dek">Install Munder Difflin on macOS, Windows, or Linux and put a hive of Claude Code agents to work on ambitious, long-horizon tasks — start to finish.</p>+ <div class="foot">+ <span class="tag kind">Guides</span>+ <a class="more" href="/blog/how-to-install-and-use-munder-difflin/" aria-label="Read: How to Install and Use Munder Difflin">Read →</a>+ </div>+ </div>+</article>+<article class="card">+ <a class="thumb t-guides" href="/blog/claude-code-automation-while-you-sleep/" aria-hidden="true" tabindex="-1">+ <span>GUIDES</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>+ </div>+ <h3><a href="/blog/claude-code-automation-while-you-sleep/">Claude Code Automation: Let Agents Build While You Sleep</a></h3>+ <p class="dek">Claude Code automation that runs overnight: an autonomous Stop-hook loop, safe permission bypass, and guardrails so you wake up to progress, not chaos.</p>+ <div class="foot">+ <span class="tag kind">Guides</span>+ <a class="more" href="/blog/claude-code-automation-while-you-sleep/" aria-label="Read: Claude Code Automation: Let Agents Build While You Sleep">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>+ <div class="body">+ <div class="meta">+ <time datetime="2026-06-04">Jun 4, 2026</time>+ <span class="dot" aria-hidden="true"></span>+ <span>7 min</span>+ </div>+ <h3><a href="/blog/context-engineering-for-ai-agents/">Context Engineering for AI Agents: Budgeting the Window</a></h3>+ <p class="dek">Why context engineering beats prompt wording for agents — and the tactics (isolation, retrieval, externalized state, compaction) that keep the window lean.</p>+ <div class="foot">+ <span class="tag kind">Guides</span>+ <a class="more" href="/blog/context-engineering-for-ai-agents/" aria-label="Read: Context Engineering for AI Agents: Budgeting the Window">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/index.html b/docs/blog/index.htmlindex 392ea83a..011f7d40 100644--- a/docs/blog/index.html+++ b/docs/blog/index.html@@ -85,9 +85,9 @@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">56</span></a>+ <a class="chip" aria-current="page">All<span class="ct">57</span></a>- <a class="chip" href="/blog/topics/guides/">guides<span class="ct">12</span></a>+ <a class="chip" href="/blog/topics/guides/">guides<span class="ct">13</span></a><a class="chip" href="/blog/topics/internals/">internals<span class="ct">10</span></a>@@ -347,6 +347,24 @@</div></div></article>+<article class="card">+ <a class="thumb t-guides" href="/blog/human-in-the-loop-approving-ai-agents/" aria-hidden="true" tabindex="-1">+ <span>GUIDES</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/human-in-the-loop-approving-ai-agents/">Approving AI Agents: Human-in-the-Loop Without a Queue</a></h3>+ <p class="dek">How to approve what AI agents do without a parallel approval queue: native permission prompts, routing 'to: human' to the orchestrator, remote approval.</p>+ <div class="foot">+ <span class="tag kind">Guides</span>+ <a class="more" href="/blog/human-in-the-loop-approving-ai-agents/" aria-label="Read: Approving AI Agents: Human-in-the-Loop Without a Queue">Read →</a>+ </div>+ </div>+</article><article class="card"><a class="thumb t-memory" href="/blog/keep-agent-semantic-memory-clean/" aria-hidden="true" tabindex="-1"><span>MEMORY</span>diff --git a/docs/blog/keep-agent-semantic-memory-clean/index.html b/docs/blog/keep-agent-semantic-memory-clean/index.htmlindex c2319314..8a238d51 100644--- a/docs/blog/keep-agent-semantic-memory-clean/index.html+++ b/docs/blog/keep-agent-semantic-memory-clean/index.html@@ -283,7 +283,7 @@ sticks; it’s free and open source.</p><footer class="post-foot wrap"><div class="prevnext"><a class="pn prev" href="/blog/local-first-ai-agent-orchestration/"><span class="k">← Older</span><span class="t">Local-First AI Agent Orchestration: How a Hive Runs Without a Cloud</span></a>- <a class="pn next" href="/blog/how-ai-agents-verify-their-own-work/"><span class="k">Newer →</span><span class="t">How AI Agents Verify Their Own Work (Before Saying 'Done')</span></a>+ <a class="pn next" href="/blog/human-in-the-loop-approving-ai-agents/"><span class="k">Newer →</span><span class="t">Approving AI Agents: Human-in-the-Loop Without a Queue</span></a></div><section class="related"><h2>Related in Memory</h2>diff --git a/docs/blog/tags/hive/index.html b/docs/blog/tags/hive/index.htmlindex 4d668a29..58bf70e6 100644--- a/docs/blog/tags/hive/index.html+++ b/docs/blog/tags/hive/index.html@@ -58,11 +58,29 @@<header class="page-head wrap"><p class="eyebrow">Tag</p><h1>#Hive</h1>- <p class="lead">1 post tagged <strong>Hive</strong>.</p>+ <p class="lead">2 posts tagged <strong>Hive</strong>.</p></header><section class="wrap"><div class="post-grid"><article class="card">+ <a class="thumb t-guides" href="/blog/human-in-the-loop-approving-ai-agents/" aria-hidden="true" tabindex="-1">+ <span>GUIDES</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/human-in-the-loop-approving-ai-agents/">Approving AI Agents: Human-in-the-Loop Without a Queue</a></h3>+ <p class="dek">How to approve what AI agents do without a parallel approval queue: native permission prompts, routing 'to: human' to the orchestrator, remote approval.</p>+ <div class="foot">+ <span class="tag kind">Guides</span>+ <a class="more" href="/blog/human-in-the-loop-approving-ai-agents/" aria-label="Read: Approving AI Agents: Human-in-the-Loop Without a Queue">Read →</a>+ </div>+ </div>+</article>+<article class="card"><a class="thumb t-orchestration" href="/blog/scheduling-autonomous-agent-missions/" aria-hidden="true" tabindex="-1"><span>ORCHESTRATION</span></a>diff --git a/docs/blog/tags/human-in-the-loop/index.html b/docs/blog/tags/human-in-the-loop/index.htmlindex 27166848..fe0b133f 100644--- a/docs/blog/tags/human-in-the-loop/index.html+++ b/docs/blog/tags/human-in-the-loop/index.html@@ -58,11 +58,29 @@<header class="page-head wrap"><p class="eyebrow">Tag</p><h1>#Human-in-the-Loop</h1>- <p class="lead">2 posts tagged <strong>Human-in-the-Loop</strong>.</p>+ <p class="lead">3 posts tagged <strong>Human-in-the-Loop</strong>.</p></header><section class="wrap"><div class="post-grid"><article class="card">+ <a class="thumb t-guides" href="/blog/human-in-the-loop-approving-ai-agents/" aria-hidden="true" tabindex="-1">+ <span>GUIDES</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/human-in-the-loop-approving-ai-agents/">Approving AI Agents: Human-in-the-Loop Without a Queue</a></h3>+ <p class="dek">How to approve what AI agents do without a parallel approval queue: native permission prompts, routing 'to: human' to the orchestrator, remote approval.</p>+ <div class="foot">+ <span class="tag kind">Guides</span>+ <a class="more" href="/blog/human-in-the-loop-approving-ai-agents/" aria-label="Read: Approving AI Agents: Human-in-the-Loop Without a Queue">Read →</a>+ </div>+ </div>+</article>+<article class="card"><a class="thumb t-use-cases" href="/blog/overnight-ai-coding-realistic-expectations/" aria-hidden="true" tabindex="-1"><span>USE CASES</span></a>diff --git a/docs/blog/tags/internals/index.html b/docs/blog/tags/internals/index.htmlindex 8c23b579..150a8076 100644--- a/docs/blog/tags/internals/index.html+++ b/docs/blog/tags/internals/index.html@@ -58,7 +58,7 @@<header class="page-head wrap"><p class="eyebrow">Tag</p><h1>#Internals</h1>- <p class="lead">12 posts tagged <strong>Internals</strong>.</p>+ <p class="lead">13 posts tagged <strong>Internals</strong>.</p></header><section class="wrap"><div class="post-grid">@@ -80,6 +80,24 @@</div></div></article>+<article class="card">+ <a class="thumb t-guides" href="/blog/human-in-the-loop-approving-ai-agents/" aria-hidden="true" tabindex="-1">+ <span>GUIDES</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/human-in-the-loop-approving-ai-agents/">Approving AI Agents: Human-in-the-Loop Without a Queue</a></h3>+ <p class="dek">How to approve what AI agents do without a parallel approval queue: native permission prompts, routing 'to: human' to the orchestrator, remote approval.</p>+ <div class="foot">+ <span class="tag kind">Guides</span>+ <a class="more" href="/blog/human-in-the-loop-approving-ai-agents/" aria-label="Read: Approving AI Agents: Human-in-the-Loop Without a Queue">Read →</a>+ </div>+ </div>+</article><article class="card"><a class="thumb t-orchestration" href="/blog/local-first-ai-agent-orchestration/" aria-hidden="true" tabindex="-1"><span>ORCHESTRATION</span>diff --git a/docs/blog/tags/orchestration/index.html b/docs/blog/tags/orchestration/index.htmlindex 3d350e8a..72149572 100644--- a/docs/blog/tags/orchestration/index.html+++ b/docs/blog/tags/orchestration/index.html@@ -58,7 +58,7 @@<header class="page-head wrap"><p class="eyebrow">Tag</p><h1>#Orchestration</h1>- <p class="lead">11 posts tagged <strong>Orchestration</strong>.</p>+ <p class="lead">12 posts tagged <strong>Orchestration</strong>.</p></header><section class="wrap"><div class="post-grid">@@ -116,6 +116,24 @@</div></div></article>+<article class="card">+ <a class="thumb t-guides" href="/blog/human-in-the-loop-approving-ai-agents/" aria-hidden="true" tabindex="-1">+ <span>GUIDES</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/human-in-the-loop-approving-ai-agents/">Approving AI Agents: Human-in-the-Loop Without a Queue</a></h3>+ <p class="dek">How to approve what AI agents do without a parallel approval queue: native permission prompts, routing 'to: human' to the orchestrator, remote approval.</p>+ <div class="foot">+ <span class="tag kind">Guides</span>+ <a class="more" href="/blog/human-in-the-loop-approving-ai-agents/" aria-label="Read: Approving AI Agents: Human-in-the-Loop Without a Queue">Read →</a>+ </div>+ </div>+</article><article class="card"><a class="thumb t-orchestration" href="/blog/local-first-ai-agent-orchestration/" aria-hidden="true" tabindex="-1"><span>ORCHESTRATION</span>diff --git a/docs/blog/topics/comparisons/index.html b/docs/blog/topics/comparisons/index.htmlindex dfedc4cd..9911af43 100644--- a/docs/blog/topics/comparisons/index.html+++ b/docs/blog/topics/comparisons/index.html@@ -68,7 +68,7 @@<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">12</span></a>+ <a class="chip" href="/blog/topics/guides/">guides<span class="ct">13</span></a><a class="chip" href="/blog/topics/internals/">internals<span class="ct">10</span></a>diff --git a/docs/blog/topics/concepts/index.html b/docs/blog/topics/concepts/index.htmlindex be890ace..858c3c69 100644--- a/docs/blog/topics/concepts/index.html+++ b/docs/blog/topics/concepts/index.html@@ -68,7 +68,7 @@<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">12</span></a>+ <a class="chip" href="/blog/topics/guides/">guides<span class="ct">13</span></a><a class="chip" href="/blog/topics/internals/">internals<span class="ct">10</span></a>diff --git a/docs/blog/topics/guides/index.html b/docs/blog/topics/guides/index.htmlindex 981f2e79..bac9cef4 100644--- a/docs/blog/topics/guides/index.html+++ b/docs/blog/topics/guides/index.html@@ -63,12 +63,12 @@</nav><p class="eyebrow">Topic</p><h1>guides</h1>- <p class="lead">12 posts in this cluster.</p>+ <p class="lead">13 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/" aria-current="page">guides<span class="ct">12</span></a>+ <a class="chip" href="/blog/topics/guides/" aria-current="page">guides<span class="ct">13</span></a><a class="chip" href="/blog/topics/internals/">internals<span class="ct">10</span></a>@@ -161,6 +161,24 @@</div></div></article>+<article class="card">+ <a class="thumb t-guides" href="/blog/human-in-the-loop-approving-ai-agents/" aria-hidden="true" tabindex="-1">+ <span>GUIDES</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/human-in-the-loop-approving-ai-agents/">Approving AI Agents: Human-in-the-Loop Without a Queue</a></h3>+ <p class="dek">How to approve what AI agents do without a parallel approval queue: native permission prompts, routing 'to: human' to the orchestrator, remote approval.</p>+ <div class="foot">+ <span class="tag kind">Guides</span>+ <a class="more" href="/blog/human-in-the-loop-approving-ai-agents/" aria-label="Read: Approving AI Agents: Human-in-the-Loop Without a Queue">Read →</a>+ </div>+ </div>+</article><article class="card"><a class="thumb t-guides" href="/blog/prompt-caching-for-ai-agents/" aria-hidden="true" tabindex="-1"><span>GUIDES</span>diff --git a/docs/blog/topics/index.html b/docs/blog/topics/index.htmlindex 1d90fd6a..2ae0aa61 100644--- a/docs/blog/topics/index.html+++ b/docs/blog/topics/index.html@@ -70,7 +70,7 @@<span class="thumb t-guides" aria-hidden="true">GUIDES</span><div class="body"><h3 style="color:var(--ink)">guides</h3>- <p class="dek">12 posts.</p>+ <p class="dek">13 posts.</p><div class="foot"><span class="more">View topic →</span></div></div></a>@@ -148,9 +148,9 @@<a class="chip" href="/blog/tags/claude-code/">Claude Code<span class="ct">31</span></a>- <a class="chip" href="/blog/tags/internals/">Internals<span class="ct">12</span></a>+ <a class="chip" href="/blog/tags/internals/">Internals<span class="ct">13</span></a>- <a class="chip" href="/blog/tags/orchestration/">Orchestration<span class="ct">11</span></a>+ <a class="chip" href="/blog/tags/orchestration/">Orchestration<span class="ct">12</span></a><a class="chip" href="/blog/tags/guides/">Guides<span class="ct">9</span></a>@@ -174,6 +174,8 @@<a class="chip" href="/blog/tags/observability/">Observability<span class="ct">3</span></a>+ <a class="chip" href="/blog/tags/human-in-the-loop/">Human-in-the-Loop<span class="ct">3</span></a>+<a class="chip" href="/blog/tags/local-first/">Local-First<span class="ct">3</span></a><a class="chip" href="/blog/tags/terminals/">Terminals<span class="ct">3</span></a>@@ -184,8 +186,6 @@<a class="chip" href="/blog/tags/god/">GOD<span class="ct">2</span></a>- <a class="chip" href="/blog/tags/human-in-the-loop/">Human-in-the-Loop<span class="ct">2</span></a>-<a class="chip" href="/blog/tags/story/">Story<span class="ct">2</span></a><a class="chip" href="/blog/tags/git/">Git<span class="ct">2</span></a>@@ -202,6 +202,8 @@<a class="chip" href="/blog/tags/cost/">Cost<span class="ct">2</span></a>+ <a class="chip" href="/blog/tags/hive/">Hive<span class="ct">2</span></a>+<a class="chip" href="/blog/tags/subagents/">Subagents<span class="ct">1</span></a><a class="chip" href="/blog/tags/markdown/">Markdown<span class="ct">1</span></a>@@ -246,8 +248,6 @@<a class="chip" href="/blog/tags/prompt-caching/">Prompt Caching<span class="ct">1</span></a>- <a class="chip" href="/blog/tags/hive/">Hive<span class="ct">1</span></a>-<a class="chip" href="/blog/tags/aeo/">AEO<span class="ct">1</span></a><a class="chip" href="/blog/tags/seo/">SEO<span class="ct">1</span></a>diff --git a/docs/blog/topics/internals/index.html b/docs/blog/topics/internals/index.htmlindex 971d2b21..0d29ae3d 100644--- a/docs/blog/topics/internals/index.html+++ b/docs/blog/topics/internals/index.html@@ -68,7 +68,7 @@<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">12</span></a>+ <a class="chip" href="/blog/topics/guides/">guides<span class="ct">13</span></a><a class="chip" href="/blog/topics/internals/" aria-current="page">internals<span class="ct">10</span></a>diff --git a/docs/blog/topics/memory/index.html b/docs/blog/topics/memory/index.htmlindex 0c7a7c9a..bf13ef58 100644--- a/docs/blog/topics/memory/index.html+++ b/docs/blog/topics/memory/index.html@@ -68,7 +68,7 @@<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">12</span></a>+ <a class="chip" href="/blog/topics/guides/">guides<span class="ct">13</span></a><a class="chip" href="/blog/topics/internals/">internals<span class="ct">10</span></a>diff --git a/docs/blog/topics/orchestration/index.html b/docs/blog/topics/orchestration/index.htmlindex 7eb8870a..b1e4d3c6 100644--- a/docs/blog/topics/orchestration/index.html+++ b/docs/blog/topics/orchestration/index.html@@ -68,7 +68,7 @@<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">12</span></a>+ <a class="chip" href="/blog/topics/guides/">guides<span class="ct">13</span></a><a class="chip" href="/blog/topics/internals/">internals<span class="ct">10</span></a>diff --git a/docs/blog/topics/story/index.html b/docs/blog/topics/story/index.htmlindex f897cd8b..5f236dd9 100644--- a/docs/blog/topics/story/index.html+++ b/docs/blog/topics/story/index.html@@ -68,7 +68,7 @@<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">12</span></a>+ <a class="chip" href="/blog/topics/guides/">guides<span class="ct">13</span></a><a class="chip" href="/blog/topics/internals/">internals<span class="ct">10</span></a>diff --git a/docs/blog/topics/use-cases/index.html b/docs/blog/topics/use-cases/index.htmlindex f9d4249f..67399905 100644--- a/docs/blog/topics/use-cases/index.html+++ b/docs/blog/topics/use-cases/index.html@@ -68,7 +68,7 @@<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">12</span></a>+ <a class="chip" href="/blog/topics/guides/">guides<span class="ct">13</span></a><a class="chip" href="/blog/topics/internals/">internals<span class="ct">10</span></a>diff --git a/docs/sitemap.xml b/docs/sitemap.xmlindex 7eb4379e..baf61158 100644--- a/docs/sitemap.xml+++ b/docs/sitemap.xml@@ -95,6 +95,12 @@<changefreq>monthly</changefreq><priority>0.8</priority></url>+ <url>+ <loc>https://munderdiffl.in/blog/human-in-the-loop-approving-ai-agents/</loc>+ <lastmod>2026-06-04</lastmod>+ <changefreq>monthly</changefreq>+ <priority>0.8</priority>+ </url><url><loc>https://munderdiffl.in/blog/keep-agent-semantic-memory-clean/</loc><lastmod>2026-06-04</lastmod>@@ -376,12 +382,12 @@<url><loc>https://munderdiffl.in/blog/tags/use-cases/</loc><changefreq>weekly</changefreq><priority>0.3</priority></url><url><loc>https://munderdiffl.in/blog/tags/workflow/</loc><changefreq>weekly</changefreq><priority>0.3</priority></url><url><loc>https://munderdiffl.in/blog/tags/observability/</loc><changefreq>weekly</changefreq><priority>0.3</priority></url>+ <url><loc>https://munderdiffl.in/blog/tags/human-in-the-loop/</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/terminals/</loc><changefreq>weekly</changefreq><priority>0.3</priority></url><url><loc>https://munderdiffl.in/blog/tags/electron/</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><url><loc>https://munderdiffl.in/blog/tags/god/</loc><changefreq>weekly</changefreq><priority>0.3</priority></url>- <url><loc>https://munderdiffl.in/blog/tags/human-in-the-loop/</loc><changefreq>weekly</changefreq><priority>0.3</priority></url><url><loc>https://munderdiffl.in/blog/tags/story/</loc><changefreq>weekly</changefreq><priority>0.3</priority></url><url><loc>https://munderdiffl.in/blog/tags/git/</loc><changefreq>weekly</changefreq><priority>0.3</priority></url><url><loc>https://munderdiffl.in/blog/tags/pixijs/</loc><changefreq>weekly</changefreq><priority>0.3</priority></url>@@ -390,6 +396,7 @@<url><loc>https://munderdiffl.in/blog/tags/xtermjs/</loc><changefreq>weekly</changefreq><priority>0.3</priority></url><url><loc>https://munderdiffl.in/blog/tags/performance/</loc><changefreq>weekly</changefreq><priority>0.3</priority></url><url><loc>https://munderdiffl.in/blog/tags/cost/</loc><changefreq>weekly</changefreq><priority>0.3</priority></url>+ <url><loc>https://munderdiffl.in/blog/tags/hive/</loc><changefreq>weekly</changefreq><priority>0.3</priority></url><url><loc>https://munderdiffl.in/blog/tags/subagents/</loc><changefreq>weekly</changefreq><priority>0.3</priority></url><url><loc>https://munderdiffl.in/blog/tags/markdown/</loc><changefreq>weekly</changefreq><priority>0.3</priority></url><url><loc>https://munderdiffl.in/blog/tags/coordination/</loc><changefreq>weekly</changefreq><priority>0.3</priority></url>@@ -412,7 +419,6 @@<url><loc>https://munderdiffl.in/blog/tags/reliability/</loc><changefreq>weekly</changefreq><priority>0.3</priority></url><url><loc>https://munderdiffl.in/blog/tags/trust/</loc><changefreq>weekly</changefreq><priority>0.3</priority></url><url><loc>https://munderdiffl.in/blog/tags/prompt-caching/</loc><changefreq>weekly</changefreq><priority>0.3</priority></url>- <url><loc>https://munderdiffl.in/blog/tags/hive/</loc><changefreq>weekly</changefreq><priority>0.3</priority></url><url><loc>https://munderdiffl.in/blog/tags/aeo/</loc><changefreq>weekly</changefreq><priority>0.3</priority></url><url><loc>https://munderdiffl.in/blog/tags/seo/</loc><changefreq>weekly</changefreq><priority>0.3</priority></url><url><loc>https://munderdiffl.in/blog/tags/devrel/</loc><changefreq>weekly</changefreq><priority>0.3</priority></url>