Compress tool outputs, logs, files, and RAG chunks before they reach the LLM. 20% fewer tokens for coding agents, 60-95% fewer tokens for JSON, same answers. Library, proxy, MCP server.
Python agentaianthropicclaude-codecompressioncontext-engineeringcontext-windowcursorfastapilangchainllmmcpopenaiprompt-engineeringproxypythonragtoken-optimizationtokenstypescript
Commits on main
fix(proxy/openai): feed Codex WS traffic into the traffic learner (#2334)
f669149 Abhay Singh committed 3 months ago
fix(proxy): pass through cross-region prefixed Bedrock model IDs directly (#2330)
64cb46e Matt Haitana committed 3 months ago
fix(proxy): preserve signed Anthropic thinking blocks on outbound re-serialize (#2254)
dc163bc AxelRay committed 3 months ago
fix(backends/litellm): None-guard core token counts in OpenAI usage block (#2324)
12f9f58 Abhay Singh committed 3 months ago
fix(transforms/adaptive-sizer): honor max_k on small-input fast path (#2319)
8a90523 Abhay Singh committed 3 months ago
fix(proxy/bedrock): report uncached input tokens from backend usage, not the live-zone count (#2318)
c19e412 Abhay Singh committed 3 months ago
fix(proxy/batch): don't crash an OpenAI batch on a valid-JSON non-object line (#2316)
1f2c681 Abhay Singh committed 3 months ago
fix(learn): stop classifying a successful exit code 0 as an error (#2289)
a24fe7d Abhay Singh committed 3 months ago
fix(learn/grok): detect a Windows absolute project path (#2283)
e240df2 Abhay Singh committed 3 months ago
fix(proxy/openai): don't crash the Responses memory tool loops on null arguments (#2273)
a30db2c Abhay Singh committed 3 months ago
fix(tokenizer): price CJK in the Rust fixed-ratio estimator (Python parity) (#2260)
6840153 Zhenjia ZHOU committed 3 months ago
fix(ci): publish latest from the root Docker manifest (#2252)
5568d73 Rod Boev committed 3 months ago
fix(transforms/smart_crusher): don't crash on a tool call with a null function (#2232)
3bb02f8 Abhay Singh committed 3 months ago
fix(memory): make explicit-project and user store keys collision-resistant (#2231)
f840d5f Abhay Singh committed 3 months ago
fix(learn/gemini): stop double-counting session tokens (#2230)
29d8a5e Abhay Singh committed 3 months ago
fix(learn/gemini): detect the project path for JSONL sessions (#2229)
7e83b8d Abhay Singh committed 3 months ago
fix(proxy): give each Codex /v1/responses WS turn a unique request_id (#2164)
d02df10 Rod Boev committed 3 months ago
fix(proxy): gate mid-turn message coalescing to Claude Code clients (#1643)
a4bd2e6 TenderDeve committed 3 months ago
fix(health): label kompress as degraded/optional when not yet loaded (#2865)
8949371 Radhakrishnan Pachyappan committed 3 months ago
fix(ccr): report embedded hashes from compress endpoint (#717)
685ebe4 Alex Sun committed 3 months ago
fix(proxy): allow settings routes for trusted gateway/dashboard clients (#2491)
a5b0a8f Parideboy committed 3 months ago
fix: Vertex model pricing shows $0.00 for versioned model names and vertex:anthropic provider (#2517)
eb5b5e4 Yossi Ovadia committed 3 months ago
fix(proxy): include tool_search_deferral savings in the savings ledger
12149f7 Abhay Singh committed 3 months ago
fix(proxy): close the upstream stream when a streaming body is never consumed
0951663 Abhay Singh committed 3 months ago
fix(content-router): protect custom-tag blocks before mixed-content section split
d7bc1e2 gglucass committed 3 months ago
fix(backends/anyllm): stream tool_use blocks and map finish_reason on the streaming path
e4904e2 Abhay Singh committed 3 months ago
fix(backends/anyllm): convert Anthropic tools and tool_choice to OpenAI shape
0d6866b Abhay Singh committed 3 months ago
fix(cache): mirror client cache_control positions instead of single-marker consolidation
def3d76 gglucass committed 3 months ago