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): kill the image worker a timed-out call abandons (#3940)
793bb85 gglucass committed 1 month ago
fix(router): fall back to built-ins when an external compressor passes through (#3893)
2c4b20f Recep Kurtulus committed 1 month ago
fix(proxy): freeze and replay the forwarded prefix on Gemini paths (#3394) (#3865)
dd84321 davidscottpope-gif committed 1 month ago
fix(sdk): preserve Gemini media parts in message conversion instead of dropping the turn (#3882)
038c923 Shubham Srivastava committed 1 month ago
fix(security): poll usage only with the operator's own credential (#3863)
65e63c0 Tejas Chopra committed 1 month ago
fix(backends/litellm): align streaming usage tokens with the non-streaming path (#2688)
90a68e0 Abhay Singh committed 1 month ago
fix(backends/litellm): map Anthropic tool_choice none to OpenAI none (#2689)
57a5708 Abhay Singh committed 1 month ago
docs(configuration): document LiteLLM as the source for limits and prices (#2781)
95ae69a Tejas Chopra committed 1 month ago
fix(oauth2): mint and inject only on requests that go upstream (#3849)
f977d52 Tejas Chopra committed 1 month ago
fix(security): stop forwarding the proxy token to upstream providers (#3891)
0147cf0 Satya Bharadwaj committed 1 month ago
fix(doctor): recognize Azure Foundry routing for Claude Code (#1339)
d5318ac jkn committed 1 month ago
fix(wrap): preserve pre-set ANTHROPIC_BASE_URL as proxy upstream (#1358)
bb1ab6a Ben Younes committed 1 month ago
fix(parser): whitespace waste signal always reported zero (#1102)
f19bc9a nyxst4ck committed 1 month ago
fix: surface Codex responses traffic in dashboard (#399)
ecc4967 Mohamed EL HAJJAJI committed 1 month ago
fix(learn): keep transcript-derived content inside the managed block (#3850)
c46e74d Tejas Chopra committed 1 month ago
fix(proxy): keep exception text and the proxy token out of client output (#3851)
861e94d Tejas Chopra committed 1 month ago
fix(proxy): make --stateless write nothing to the workspace (#3846)
3c9ed01 Tejas Chopra committed 1 month ago
fix(backends/litellm): report requested model in OpenAI streaming chunks (#2690)
8538a83 Abhay Singh committed 1 month ago
refactor(ccr): read the expansion query through extract_user_query (#2707)
dcec845 nangsontay committed 1 month ago
test(proxy): assert bounded concurrency structurally in the pre-upstream gate (#2831)
885af38 Parideboy committed 1 month ago
fix(toin): honor HEADROOM_TOIN_BACKEND=none instead of falling back to disk (#2884)
9a9f35c probe committed 1 month ago
fix(image): cap the technique-router ONNX intra-op thread pool (#2885)
d7ed230 probe committed 1 month ago
fix(proxy): preserve Windows service when Rust core is blocked (#2989)
eaa16d9 Parideboy committed 1 month ago
feat(memory): persist TrafficLearner pending evidence across restarts + expose learner stats (#3104)
4257ed4 gglucass committed 1 month ago
fix(utils): normalize timezone-aware datetimes in format_timestamp (#3116)
b625d18 Abhay Singh committed 1 month ago
fix(copilot): read timezone-naive token expiry as UTC, not host local time (#3216)
2b2dc1b Rajat Nagda committed 1 month ago