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
deps: bump ruff from 0.15.22 to 0.16.2 in the pip-minor-patch group across 1 directory (#2962)
dependabot[bot]ff17961cd76a
diff --git a/.pre-commit-config.yaml b/.pre-commit-config.yamlindex 08ee4e215..58d46b666 100644--- a/.pre-commit-config.yaml+++ b/.pre-commit-config.yaml@@ -27,7 +27,7 @@ repos:# unconditionally, so installing hooks is not required for enforcement.args: [--assume-in-merge]- repo: https://github.com/astral-sh/ruff-pre-commit- rev: v0.15.22+ rev: v0.16.2hooks:- id: ruffargs: [--fix]diff --git a/headroom/testing/README.md b/headroom/testing/README.mdindex 7b9ee7f99..37c398669 100644--- a/headroom/testing/README.md+++ b/headroom/testing/README.md@@ -33,7 +33,11 @@ assert report.passedsuite = (Headroom.Suite("phase-1").Add(Headroom.WithOpenAI().named("openai-cache").WithCompression(mode="cache"))- .Add(Headroom.WithBedrock(region="us-east-1").named("bedrock-token").WithCompression(mode="token"))+ .Add(+ Headroom.WithBedrock(region="us-east-1")+ .named("bedrock-token")+ .WithCompression(mode="token")+ ))suite.write_manifest_bundle("headroom-testing-bundle.json", provider="openai", port_start=19000)diff --git a/pyproject.toml b/pyproject.tomlindex e36dca4ae..c4450a1e4 100644--- a/pyproject.toml+++ b/pyproject.toml@@ -277,7 +277,7 @@ dev = ["pytest>=7.0.0","pytest-cov>=4.0.0","pytest-asyncio>=0.21.0",- "ruff==0.15.22",+ "ruff==0.16.2","mypy>=1.0.0","pre-commit>=3.0.0","openai>=1.0.0",diff --git a/uv.lock b/uv.lockindex edd0b3646..45feb1dd2 100644--- a/uv.lock+++ b/uv.lock@@ -280,12 +280,12 @@ name = "any-llm-sdk"version = "1.12.1"source = { registry = "https://pypi.org/simple/" }dependencies = [- { name = "httpx" },- { name = "openai" },- { name = "openresponses-types" },- { name = "pydantic" },- { name = "rich" },- { name = "typing-extensions" },+ { name = "httpx", marker = "python_full_version >= '3.11'" },+ { name = "openai", marker = "python_full_version >= '3.11'" },+ { name = "openresponses-types", marker = "python_full_version >= '3.11'" },+ { name = "pydantic", marker = "python_full_version >= '3.11'" },+ { name = "rich", marker = "python_full_version >= '3.11'" },+ { name = "typing-extensions", marker = "python_full_version >= '3.11'" },]sdist = { url = "https://files.pythonhosted.org/packages/21/d5/21ef27d031b72f0054ba0015a66cc1f10cda9410e4a3acb6e795d848ad0d/any_llm_sdk-1.12.1.tar.gz", hash = "sha256:76e043fcaa56fccfb375a908511869dc7dbf393dd97a82d06bb764d8201724e8", size = 152078, upload-time = "2026-03-18T13:13:01.735Z" }wheels = [@@ -811,7 +811,7 @@ name = "coloredlogs"version = "15.0.1"source = { registry = "https://pypi.org/simple/" }dependencies = [- { name = "humanfriendly" },+ { name = "humanfriendly", marker = "python_full_version < '3.11'" },]sdist = { url = "https://files.pythonhosted.org/packages/cc/c7/eed8f27100517e8c0e6b923d5f0845d0cb99763da6fdee00478f91db7325/coloredlogs-15.0.1.tar.gz", hash = "sha256:7c991aa71a4577af2f82600d8f8f3a89f936baeaf9b50a9c197da014e5bf16b0", size = 278520, upload-time = "2021-06-11T10:22:45.202Z" }wheels = [@@ -823,7 +823,7 @@ name = "colorlog"version = "6.10.1"source = { registry = "https://pypi.org/simple/" }dependencies = [- { name = "colorama", marker = "sys_platform == 'win32'" },+ { name = "colorama", marker = "python_full_version >= '3.13' and sys_platform == 'win32'" },]sdist = { url = "https://files.pythonhosted.org/packages/a2/61/f083b5ac52e505dfc1c624eafbf8c7589a0d7f32daa398d2e7590efa5fda/colorlog-6.10.1.tar.gz", hash = "sha256:eb4ae5cb65fe7fec7773c2306061a8e63e02efc2c72eba9d27b0fa23c94f1321", size = 17162, upload-time = "2025-10-16T16:14:11.978Z" }wheels = [@@ -1045,7 +1045,7 @@ name = "cuda-bindings"version = "13.3.1"source = { registry = "https://pypi.org/simple/" }dependencies = [- { name = "cuda-pathfinder" },+ { name = "cuda-pathfinder", marker = "(python_full_version < '3.11' and sys_platform == 'emscripten') or (python_full_version < '3.11' and sys_platform == 'win32') or (sys_platform != 'emscripten' and sys_platform != 'win32')" },]wheels = [{ url = "https://files.pythonhosted.org/packages/a9/21/8464d133752951c154feafb3b65c297e7d80f301183d220bec4c830f1441/cuda_bindings-13.3.1-cp310-cp310-manylinux_2_24_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:120fcc53d57903df529c3486962c56528cba5b7d6c57c99537320ed9922c8b86", size = 6073403, upload-time = "2026-05-29T23:11:36.22Z" },@@ -1080,43 +1080,43 @@ wheels = [[package.optional-dependencies]cublas = [- { name = "nvidia-cublas", marker = "platform_machine == 'aarch64' or platform_machine == 'x86_64'" },- { name = "nvidia-cuda-nvrtc", marker = "platform_machine == 'aarch64' or platform_machine == 'x86_64'" },+ { name = "nvidia-cublas", marker = "(python_full_version < '3.11' and platform_machine == 'AMD64' and sys_platform == 'win32') or (platform_machine == 'aarch64' and sys_platform == 'linux') or (platform_machine == 'x86_64' and sys_platform == 'linux')" },+ { name = "nvidia-cuda-nvrtc", marker = "(python_full_version < '3.11' and platform_machine == 'AMD64' and sys_platform == 'win32') or (platform_machine == 'aarch64' and sys_platform == 'linux') or (platform_machine == 'x86_64' and sys_platform == 'linux')" },]cudart = [- { name = "nvidia-cuda-runtime", marker = "platform_machine == 'aarch64' or platform_machine == 'x86_64'" },+ { name = "nvidia-cuda-runtime", marker = "(python_full_version < '3.11' and platform_machine == 'AMD64' and sys_platform == 'win32') or (platform_machine == 'aarch64' and sys_platform == 'linux') or (platform_machine == 'x86_64' and sys_platform == 'linux')" },]cufft = [- { name = "nvidia-cufft", marker = "platform_machine == 'aarch64' or platform_machine == 'x86_64'" },- { name = "nvidia-nvjitlink", marker = "platform_machine == 'aarch64' or platform_machine == 'x86_64'" },+ { name = "nvidia-cufft", marker = "(python_full_version < '3.11' and platform_machine == 'AMD64' and sys_platform == 'win32') or (platform_machine == 'aarch64' and sys_platform == 'linux') or (platform_machine == 'x86_64' and sys_platform == 'linux')" },+ { name = "nvidia-nvjitlink", marker = "(python_full_version < '3.11' and platform_machine == 'AMD64' and sys_platform == 'win32') or (platform_machine == 'aarch64' and sys_platform == 'linux') or (platform_machine == 'x86_64' and sys_platform == 'linux')" },]cufile = [- { name = "nvidia-cufile", marker = "platform_machine == 'aarch64' or platform_machine == 'x86_64'" },+ { name = "nvidia-cufile", marker = "(platform_machine == 'aarch64' and sys_platform == 'linux') or (platform_machine == 'x86_64' and sys_platform == 'linux')" },]cupti = [- { name = "nvidia-cuda-cupti", marker = "platform_machine == 'aarch64' or platform_machine == 'x86_64'" },+ { name = "nvidia-cuda-cupti", marker = "(python_full_version < '3.11' and platform_machine == 'AMD64' and sys_platform == 'win32') or (platform_machine == 'aarch64' and sys_platform == 'linux') or (platform_machine == 'x86_64' and sys_platform == 'linux')" },]curand = [- { name = "nvidia-curand", marker = "platform_machine == 'aarch64' or platform_machine == 'x86_64'" },+ { name = "nvidia-curand", marker = "(python_full_version < '3.11' and platform_machine == 'AMD64' and sys_platform == 'win32') or (platform_machine == 'aarch64' and sys_platform == 'linux') or (platform_machine == 'x86_64' and sys_platform == 'linux')" },]cusolver = [- { name = "nvidia-cublas", marker = "platform_machine == 'aarch64' or platform_machine == 'x86_64'" },- { name = "nvidia-cusolver", marker = "platform_machine == 'aarch64' or platform_machine == 'x86_64'" },- { name = "nvidia-cusparse", marker = "platform_machine == 'aarch64' or platform_machine == 'x86_64'" },- { name = "nvidia-nvjitlink", marker = "platform_machine == 'aarch64' or platform_machine == 'x86_64'" },+ { name = "nvidia-cublas", marker = "(python_full_version < '3.11' and platform_machine == 'AMD64' and sys_platform == 'win32') or (platform_machine == 'aarch64' and sys_platform == 'linux') or (platform_machine == 'x86_64' and sys_platform == 'linux')" },+ { name = "nvidia-cusolver", marker = "(python_full_version < '3.11' and platform_machine == 'AMD64' and sys_platform == 'win32') or (platform_machine == 'aarch64' and sys_platform == 'linux') or (platform_machine == 'x86_64' and sys_platform == 'linux')" },+ { name = "nvidia-cusparse", marker = "(python_full_version < '3.11' and platform_machine == 'AMD64' and sys_platform == 'win32') or (platform_machine == 'aarch64' and sys_platform == 'linux') or (platform_machine == 'x86_64' and sys_platform == 'linux')" },+ { name = "nvidia-nvjitlink", marker = "(python_full_version < '3.11' and platform_machine == 'AMD64' and sys_platform == 'win32') or (platform_machine == 'aarch64' and sys_platform == 'linux') or (platform_machine == 'x86_64' and sys_platform == 'linux')" },]cusparse = [- { name = "nvidia-cusparse", marker = "platform_machine == 'aarch64' or platform_machine == 'x86_64'" },- { name = "nvidia-nvjitlink", marker = "platform_machine == 'aarch64' or platform_machine == 'x86_64'" },+ { name = "nvidia-cusparse", marker = "(python_full_version < '3.11' and platform_machine == 'AMD64' and sys_platform == 'win32') or (platform_machine == 'aarch64' and sys_platform == 'linux') or (platform_machine == 'x86_64' and sys_platform == 'linux')" },+ { name = "nvidia-nvjitlink", marker = "(python_full_version < '3.11' and platform_machine == 'AMD64' and sys_platform == 'win32') or (platform_machine == 'aarch64' and sys_platform == 'linux') or (platform_machine == 'x86_64' and sys_platform == 'linux')" },]nvjitlink = [- { name = "nvidia-nvjitlink", marker = "platform_machine == 'aarch64' or platform_machine == 'x86_64'" },+ { name = "nvidia-nvjitlink", marker = "(python_full_version < '3.11' and platform_machine == 'AMD64' and sys_platform == 'win32') or (platform_machine == 'aarch64' and sys_platform == 'linux') or (platform_machine == 'x86_64' and sys_platform == 'linux')" },]nvrtc = [- { name = "nvidia-cuda-nvrtc", marker = "platform_machine == 'aarch64' or platform_machine == 'x86_64'" },+ { name = "nvidia-cuda-nvrtc", marker = "(python_full_version < '3.11' and platform_machine == 'AMD64' and sys_platform == 'win32') or (platform_machine == 'aarch64' and sys_platform == 'linux') or (platform_machine == 'x86_64' and sys_platform == 'linux')" },]nvtx = [- { name = "nvidia-nvtx", marker = "platform_machine == 'aarch64' or platform_machine == 'x86_64'" },+ { name = "nvidia-nvtx", marker = "(python_full_version < '3.11' and platform_machine == 'AMD64' and sys_platform == 'win32') or (platform_machine == 'aarch64' and sys_platform == 'linux') or (platform_machine == 'x86_64' and sys_platform == 'linux')" },][[package]]@@ -1220,7 +1220,7 @@ name = "exceptiongroup"version = "1.3.1"source = { registry = "https://pypi.org/simple/" }dependencies = [- { name = "typing-extensions" },+ { name = "typing-extensions", marker = "python_full_version < '3.11'" },]sdist = { url = "https://files.pythonhosted.org/packages/50/79/66800aadf48771f6b62f7eb014e352e5d06856655206165d775e675a02c9/exceptiongroup-1.3.1.tar.gz", hash = "sha256:8b412432c6055b0b7d14c310000ae93352ed6754f70fa8f7c34141f91c4e3219", size = 30371, upload-time = "2025-11-21T23:01:54.787Z" }wheels = [@@ -1637,7 +1637,7 @@ name = "gunicorn"version = "26.0.0"source = { registry = "https://pypi.org/simple/" }dependencies = [- { name = "packaging" },+ { name = "packaging", marker = "python_full_version < '3.11' or sys_platform != 'win32'" },]sdist = { url = "https://files.pythonhosted.org/packages/6d/b7/a4a3f632f823e432ce6bc65f62961b7980c898c77f075a2f7118cb3846fe/gunicorn-26.0.0.tar.gz", hash = "sha256:ca9346f85e3a4aeeb64d491045c16b9a35647abd37ea15efe53080eb8b090baf", size = 727286, upload-time = "2026-05-05T06:38:25.529Z" }wheels = [@@ -1988,7 +1988,7 @@ requires-dist = [{ name = "rapidocr-onnxruntime", marker = "python_full_version < '3.13' and extra == 'image'", specifier = ">=1.4.0,<2" },{ name = "respx", marker = "extra == 'dev'", specifier = ">=0.20.0" },{ name = "rich", specifier = ">=13.0.0" },- { name = "ruff", marker = "extra == 'dev'", specifier = "==0.15.22" },+ { name = "ruff", marker = "extra == 'dev'", specifier = "==0.16.2" },{ name = "scikit-learn", marker = "extra == 'evals'", specifier = ">=1.3.0" },{ name = "sentence-transformers", marker = "sys_platform == 'darwin' and extra == 'pytorch-mps'", specifier = ">=2.2.0" },{ name = "sentence-transformers", marker = "(platform_machine != 'x86_64' and extra == 'evals') or (sys_platform != 'darwin' and extra == 'evals')", specifier = ">=2.2.0,<6.0" },@@ -2207,7 +2207,7 @@ name = "humanfriendly"version = "10.0"source = { registry = "https://pypi.org/simple/" }dependencies = [- { name = "pyreadline3", marker = "sys_platform == 'win32'" },+ { name = "pyreadline3", marker = "python_full_version < '3.11' and sys_platform == 'win32'" },]sdist = { url = "https://files.pythonhosted.org/packages/cc/3f/2c29224acb2e2df4d2046e4c73ee2662023c58ff5b113c4c1adac0886c43/humanfriendly-10.0.tar.gz", hash = "sha256:6b0b831ce8f15f7300721aa49829fc4e83921a9a301cc7f606be6686a2288ddc", size = 360702, upload-time = "2021-09-17T21:40:43.31Z" }wheels = [@@ -2690,18 +2690,18 @@ name = "litellm"version = "1.88.1"source = { registry = "https://pypi.org/simple/" }dependencies = [- { name = "aiohttp" },- { name = "click" },- { name = "fastuuid" },- { name = "httpx" },- { name = "importlib-metadata" },- { name = "jinja2" },- { name = "jsonschema" },- { name = "openai" },- { name = "pydantic" },- { name = "python-dotenv" },- { name = "tiktoken" },- { name = "tokenizers" },+ { name = "aiohttp", marker = "python_full_version < '3.14'" },+ { name = "click", marker = "python_full_version < '3.14'" },+ { name = "fastuuid", marker = "python_full_version < '3.14'" },+ { name = "httpx", marker = "python_full_version < '3.14'" },+ { name = "importlib-metadata", marker = "python_full_version < '3.14'" },+ { name = "jinja2", marker = "python_full_version < '3.14'" },+ { name = "jsonschema", marker = "python_full_version < '3.14'" },+ { name = "openai", marker = "python_full_version < '3.14'" },+ { name = "pydantic", marker = "python_full_version < '3.14'" },+ { name = "python-dotenv", marker = "python_full_version < '3.14'" },+ { name = "tiktoken", marker = "python_full_version < '3.14'" },+ { name = "tokenizers", marker = "python_full_version < '3.14'" },]sdist = { url = "https://files.pythonhosted.org/packages/16/ea/f99ececb7f22703fe120f1d8be9ffb749ec9453fbbbbbebc0d6a6b4d7864/litellm-1.88.1.tar.gz", hash = "sha256:89c6b74cc7912d6365793006ff951c0450fe847625008dfe49de8a7dc4529aa5", size = 13885969, upload-time = "2026-06-09T01:06:25.192Z" }wheels = [@@ -3583,7 +3583,7 @@ name = "nvidia-cublas"version = "13.1.1.3"source = { registry = "https://pypi.org/simple/" }dependencies = [- { name = "nvidia-cuda-nvrtc" },+ { name = "nvidia-cuda-nvrtc", marker = "(python_full_version < '3.11' and sys_platform == 'emscripten') or (python_full_version < '3.11' and sys_platform == 'win32') or (sys_platform != 'emscripten' and sys_platform != 'win32')" },]wheels = [{ url = "https://files.pythonhosted.org/packages/a7/a1/0bd24ee8c8d03adac032fd2909426a00c88f8c57961b1277ded97f91119f/nvidia_cublas-13.1.1.3-py3-none-manylinux_2_27_aarch64.whl", hash = "sha256:b7a210458267ac818974c53038fbec2e969d5c99f305ab15c72522fa9f001dd5", size = 542848918, upload-time = "2026-04-08T18:46:22.985Z" },@@ -3622,7 +3622,7 @@ name = "nvidia-cudnn-cu13"version = "9.20.0.48"source = { registry = "https://pypi.org/simple/" }dependencies = [- { name = "nvidia-cublas" },+ { name = "nvidia-cublas", marker = "(python_full_version < '3.11' and sys_platform == 'emscripten') or (python_full_version < '3.11' and sys_platform == 'win32') or (sys_platform != 'emscripten' and sys_platform != 'win32')" },]wheels = [{ url = "https://files.pythonhosted.org/packages/56/c5/83384d846b2fd17c44bd499b36c75a45ed4f095fbbb2252294e89cea5c5c/nvidia_cudnn_cu13-9.20.0.48-py3-none-manylinux_2_27_aarch64.whl", hash = "sha256:e31454ae00094b0c55319d9d15b6fa2fc50a9e1c0f5c8c80fb75258234e731e1", size = 444574296, upload-time = "2026-03-09T19:28:27.751Z" },@@ -3634,7 +3634,7 @@ name = "nvidia-cufft"version = "12.0.0.61"source = { registry = "https://pypi.org/simple/" }dependencies = [- { name = "nvidia-nvjitlink" },+ { name = "nvidia-nvjitlink", marker = "(python_full_version < '3.11' and sys_platform == 'emscripten') or (python_full_version < '3.11' and sys_platform == 'win32') or (sys_platform != 'emscripten' and sys_platform != 'win32')" },]wheels = [{ url = "https://files.pythonhosted.org/packages/8b/ae/f417a75c0259e85c1d2f83ca4e960289a5f814ed0cea74d18c353d3e989d/nvidia_cufft-12.0.0.61-py3-none-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:2708c852ef8cd89d1d2068bdbece0aa188813a0c934db3779b9b1faa8442e5f5", size = 214053554, upload-time = "2025-09-04T08:31:38.196Z" },@@ -3664,9 +3664,9 @@ name = "nvidia-cusolver"version = "12.0.4.66"source = { registry = "https://pypi.org/simple/" }dependencies = [- { name = "nvidia-cublas" },- { name = "nvidia-cusparse" },- { name = "nvidia-nvjitlink" },+ { name = "nvidia-cublas", marker = "(python_full_version < '3.11' and sys_platform == 'emscripten') or (python_full_version < '3.11' and sys_platform == 'win32') or (sys_platform != 'emscripten' and sys_platform != 'win32')" },+ { name = "nvidia-cusparse", marker = "(python_full_version < '3.11' and sys_platform == 'emscripten') or (python_full_version < '3.11' and sys_platform == 'win32') or (sys_platform != 'emscripten' and sys_platform != 'win32')" },+ { name = "nvidia-nvjitlink", marker = "(python_full_version < '3.11' and sys_platform == 'emscripten') or (python_full_version < '3.11' and sys_platform == 'win32') or (sys_platform != 'emscripten' and sys_platform != 'win32')" },]wheels = [{ url = "https://files.pythonhosted.org/packages/c8/c3/b30c9e935fc01e3da443ec0116ed1b2a009bb867f5324d3f2d7e533e776b/nvidia_cusolver-12.0.4.66-py3-none-manylinux_2_27_aarch64.whl", hash = "sha256:02c2457eaa9e39de20f880f4bd8820e6a1cfb9f9a34f820eb12a155aa5bc92d2", size = 223467760, upload-time = "2025-09-04T08:33:04.222Z" },@@ -3678,7 +3678,7 @@ name = "nvidia-cusparse"version = "12.6.3.3"source = { registry = "https://pypi.org/simple/" }dependencies = [- { name = "nvidia-nvjitlink" },+ { name = "nvidia-nvjitlink", marker = "(python_full_version < '3.11' and sys_platform == 'emscripten') or (python_full_version < '3.11' and sys_platform == 'win32') or (sys_platform != 'emscripten' and sys_platform != 'win32')" },]wheels = [{ url = "https://files.pythonhosted.org/packages/f8/94/5c26f33738ae35276672f12615a64bd008ed5be6d1ebcb23579285d960a9/nvidia_cusparse-12.6.3.3-py3-none-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:80bcc4662f23f1054ee334a15c72b8940402975e0eab63178fc7e670aa59472c", size = 162155568, upload-time = "2025-09-04T08:33:42.864Z" },@@ -3757,8 +3757,8 @@ name = "omegaconf"version = "2.3.0"source = { registry = "https://pypi.org/simple/" }dependencies = [- { name = "antlr4-python3-runtime" },- { name = "pyyaml" },+ { name = "antlr4-python3-runtime", marker = "python_full_version >= '3.13'" },+ { name = "pyyaml", marker = "python_full_version >= '3.13'" },]sdist = { url = "https://files.pythonhosted.org/packages/09/48/6388f1bb9da707110532cb70ec4d2822858ddfb44f1cdf1233c20a80ea4b/omegaconf-2.3.0.tar.gz", hash = "sha256:d5d4b6d29955cc50ad50c46dc269bcd92c6e00f5f90d23ab5fee7bfca4ba4cc7", size = 3298120, upload-time = "2022-12-08T20:59:22.753Z" }wheels = [@@ -3773,12 +3773,12 @@ resolution-markers = ["python_full_version < '3.11'",]dependencies = [- { name = "coloredlogs" },- { name = "flatbuffers" },- { name = "numpy", version = "2.2.6", source = { registry = "https://pypi.org/simple/" } },- { name = "packaging" },- { name = "protobuf" },- { name = "sympy" },+ { name = "coloredlogs", marker = "python_full_version < '3.11'" },+ { name = "flatbuffers", marker = "python_full_version < '3.11'" },+ { name = "numpy", version = "2.2.6", source = { registry = "https://pypi.org/simple/" }, marker = "python_full_version < '3.11'" },+ { name = "packaging", marker = "python_full_version < '3.11'" },+ { name = "protobuf", marker = "python_full_version < '3.11'" },+ { name = "sympy", marker = "python_full_version < '3.11'" },]wheels = [{ url = "https://files.pythonhosted.org/packages/35/d6/311b1afea060015b56c742f3531168c1644650767f27ef40062569960587/onnxruntime-1.23.2-cp310-cp310-macosx_13_0_arm64.whl", hash = "sha256:a7730122afe186a784660f6ec5807138bf9d792fa1df76556b27307ea9ebcbe3", size = 17195934, upload-time = "2025-10-27T23:06:14.143Z" },@@ -3824,10 +3824,10 @@ resolution-markers = ["python_full_version == '3.11.*' and sys_platform != 'emscripten' and sys_platform != 'win32'",]dependencies = [- { name = "flatbuffers" },- { name = "numpy", version = "2.4.1", source = { registry = "https://pypi.org/simple/" } },- { name = "packaging" },- { name = "protobuf" },+ { name = "flatbuffers", marker = "python_full_version >= '3.11'" },+ { name = "numpy", version = "2.4.1", source = { registry = "https://pypi.org/simple/" }, marker = "python_full_version >= '3.11'" },+ { name = "packaging", marker = "python_full_version >= '3.11'" },+ { name = "protobuf", marker = "python_full_version >= '3.11'" },]wheels = [{ url = "https://files.pythonhosted.org/packages/d4/81/29a9eb470994a75eb7b3ccf32be314d7c66675a00ac7b50294816cc2db27/onnxruntime-1.26.0-cp311-cp311-macosx_14_0_arm64.whl", hash = "sha256:ee1109ef4ef27cad90e823399e61e03b3c6c7bfe0fb820b4baf3678c15be8b3c", size = 18005108, upload-time = "2026-05-08T19:08:11.728Z" },@@ -3911,7 +3911,7 @@ name = "openresponses-types"version = "2.3.0.post1"source = { registry = "https://pypi.org/simple/" }dependencies = [- { name = "pydantic" },+ { name = "pydantic", marker = "python_full_version >= '3.11'" },]sdist = { url = "https://files.pythonhosted.org/packages/d9/26/b612c3215f5599714fa94d63eb5ee59b4eb66dbdeeaf86bb4d848359484d/openresponses_types-2.3.0.post1.tar.gz", hash = "sha256:11b8896d3621d2ac2439f6ff106f34ddcb1bbd517c317a6c852a9df2e98a0753", size = 19254, upload-time = "2026-01-22T20:02:03.933Z" }wheels = [@@ -4154,10 +4154,10 @@ resolution-markers = ["python_full_version < '3.11'",]dependencies = [- { name = "numpy", version = "2.2.6", source = { registry = "https://pypi.org/simple/" } },- { name = "python-dateutil" },- { name = "pytz" },- { name = "tzdata" },+ { name = "numpy", version = "2.2.6", source = { registry = "https://pypi.org/simple/" }, marker = "python_full_version < '3.11'" },+ { name = "python-dateutil", marker = "python_full_version < '3.11'" },+ { name = "pytz", marker = "python_full_version < '3.11'" },+ { name = "tzdata", marker = "python_full_version < '3.11'" },]sdist = { url = "https://files.pythonhosted.org/packages/33/01/d40b85317f86cf08d853a4f495195c73815fdf205eef3993821720274518/pandas-2.3.3.tar.gz", hash = "sha256:e05e1af93b977f7eafa636d043f9f94c7ee3ac81af99c13508215942e64c993b", size = 4495223, upload-time = "2025-09-29T23:34:51.853Z" }wheels = [@@ -4229,9 +4229,9 @@ resolution-markers = ["python_full_version == '3.11.*' and sys_platform != 'emscripten' and sys_platform != 'win32'",]dependencies = [- { name = "numpy", version = "2.4.1", source = { registry = "https://pypi.org/simple/" } },- { name = "python-dateutil" },- { name = "tzdata", marker = "sys_platform == 'emscripten' or sys_platform == 'win32'" },+ { name = "numpy", version = "2.4.1", source = { registry = "https://pypi.org/simple/" }, marker = "python_full_version >= '3.11'" },+ { name = "python-dateutil", marker = "python_full_version >= '3.11'" },+ { name = "tzdata", marker = "(python_full_version >= '3.11' and sys_platform == 'emscripten') or (python_full_version >= '3.11' and sys_platform == 'win32')" },]sdist = { url = "https://files.pythonhosted.org/packages/de/da/b1dc0481ab8d55d0f46e343cfe67d4551a0e14fcee52bd38ca1bd73258d8/pandas-3.0.0.tar.gz", hash = "sha256:0facf7e87d38f721f0af46fe70d97373a37701b1c09f7ed7aeeb292ade5c050f", size = 4633005, upload-time = "2026-01-21T15:52:04.726Z" }wheels = [@@ -5393,17 +5393,17 @@ name = "rapidocr"version = "3.8.1"source = { registry = "https://pypi.org/simple/" }dependencies = [- { name = "colorlog" },- { name = "numpy", version = "2.4.1", source = { registry = "https://pypi.org/simple/" } },- { name = "omegaconf" },- { name = "opencv-python" },- { name = "pillow" },- { name = "pyclipper" },- { name = "pyyaml" },- { name = "requests" },- { name = "shapely" },- { name = "six" },- { name = "tqdm" },+ { name = "colorlog", marker = "python_full_version >= '3.13'" },+ { name = "numpy", version = "2.4.1", source = { registry = "https://pypi.org/simple/" }, marker = "python_full_version >= '3.13'" },+ { name = "omegaconf", marker = "python_full_version >= '3.13'" },+ { name = "opencv-python", marker = "python_full_version >= '3.13'" },+ { name = "pillow", marker = "python_full_version >= '3.13'" },+ { name = "pyclipper", marker = "python_full_version >= '3.13'" },+ { name = "pyyaml", marker = "python_full_version >= '3.13'" },+ { name = "requests", marker = "python_full_version >= '3.13'" },+ { name = "shapely", marker = "python_full_version >= '3.13'" },+ { name = "six", marker = "python_full_version >= '3.13'" },+ { name = "tqdm", marker = "python_full_version >= '3.13'" },]wheels = [{ url = "https://files.pythonhosted.org/packages/ea/4a/fa521d947f0fc7bb304bf11bec4cb66266bd81494588b4cb48dc01001719/rapidocr-3.8.1-py3-none-any.whl", hash = "sha256:650044b1fbce9e6bae5cae462dcf8be754cde11e2f23fc51f65dcc08deae2c46", size = 15080319, upload-time = "2026-04-11T07:13:22.56Z" },@@ -5414,17 +5414,17 @@ name = "rapidocr-onnxruntime"version = "1.4.4"source = { registry = "https://pypi.org/simple/" }dependencies = [- { name = "numpy", version = "2.2.6", source = { registry = "https://pypi.org/simple/" }, marker = "python_full_version < '3.11' or python_full_version >= '3.13'" },+ { name = "numpy", version = "2.2.6", source = { registry = "https://pypi.org/simple/" }, marker = "python_full_version < '3.11'" },{ name = "numpy", version = "2.4.1", source = { registry = "https://pypi.org/simple/" }, marker = "python_full_version >= '3.11' and python_full_version < '3.13'" },- { name = "onnxruntime", version = "1.23.2", source = { registry = "https://pypi.org/simple/" }, marker = "python_full_version < '3.11' or python_full_version >= '3.13'" },+ { name = "onnxruntime", version = "1.23.2", source = { registry = "https://pypi.org/simple/" }, marker = "python_full_version < '3.11'" },{ name = "onnxruntime", version = "1.26.0", source = { registry = "https://pypi.org/simple/" }, marker = "python_full_version >= '3.11' and python_full_version < '3.13'" },- { name = "opencv-python" },- { name = "pillow" },- { name = "pyclipper" },- { name = "pyyaml" },- { name = "shapely" },- { name = "six" },- { name = "tqdm" },+ { name = "opencv-python", marker = "python_full_version < '3.13'" },+ { name = "pillow", marker = "python_full_version < '3.13'" },+ { name = "pyclipper", marker = "python_full_version < '3.13'" },+ { name = "pyyaml", marker = "python_full_version < '3.13'" },+ { name = "shapely", marker = "python_full_version < '3.13'" },+ { name = "six", marker = "python_full_version < '3.13'" },+ { name = "tqdm", marker = "python_full_version < '3.13'" },]wheels = [{ url = "https://files.pythonhosted.org/packages/ba/12/1e5497183bdbe782dbb91bad1d0d2297dba4d2831b2652657f7517bfc6df/rapidocr_onnxruntime-1.4.4-py3-none-any.whl", hash = "sha256:971d7d5f223a7a808662229df1ef69893809d8457d834e6373d3854bc1782cbf", size = 14915192, upload-time = "2025-01-17T01:48:25.104Z" },@@ -5754,27 +5754,27 @@ wheels = [[[package]]name = "ruff"-version = "0.15.22"-source = { registry = "https://pypi.org/simple/" }-sdist = { url = "https://files.pythonhosted.org/packages/3a/06/ae069393fc66e8ff33036d4b368003833bf6e88ccf182e17e7a2f1c754fd/ruff-0.15.22.tar.gz", hash = "sha256:3f15175b1fb580126f58285a5dae6b2ea89000136d980c64499211f116b54809", size = 4785063, upload-time = "2026-07-16T15:14:13.244Z" }-wheels = [- { url = "https://files.pythonhosted.org/packages/23/18/ee54b7ae1e121be7a28ea6da4b67564ebb0530e183a54415ab7e3bcd2c4e/ruff-0.15.22-py3-none-linux_armv6l.whl", hash = "sha256:44423e73493737f5e7c5b41d475483898ff37afcdae38bc3da5085e29af1c2d8", size = 10781258, upload-time = "2026-07-16T15:13:19.452Z" },- { url = "https://files.pythonhosted.org/packages/2f/d2/2520cb14761ddbeaf57642a76942fc36adcbdbe53b4532241995f6fc485c/ruff-0.15.22-py3-none-macosx_10_12_x86_64.whl", hash = "sha256:b82c6482946e9eda7ff2e091d25b8bad3f718684e1916d41bd56873cee05b697", size = 10999477, upload-time = "2026-07-16T15:13:23.318Z" },- { url = "https://files.pythonhosted.org/packages/c9/10/74e53572aa758dfaa678c2a2646b5c5515d884b7ca56be4d2ce03ca4b560/ruff-0.15.22-py3-none-macosx_11_0_arm64.whl", hash = "sha256:11c1c715af53a09f714e011106bffc419751ec8232fcb5da42173284ea3fec6f", size = 10466716, upload-time = "2026-07-16T15:13:26.162Z" },- { url = "https://files.pythonhosted.org/packages/1e/cc/44eaaf0844e028182f2d0a8f2190d0f359159aed0a9e5ab861d892f1ae2a/ruff-0.15.22-py3-none-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:742a29cf29bddb7c8327895d6a10e0e6c5b38a96dd407af9b5d0857f809c0576", size = 10892644, upload-time = "2026-07-16T15:13:29.229Z" },- { url = "https://files.pythonhosted.org/packages/9f/21/8edf559014d2b0f82beea19cfb713993ad802ccda16868769979c6090a84/ruff-0.15.22-py3-none-manylinux_2_17_armv7l.manylinux2014_armv7l.whl", hash = "sha256:72af58b951b0ae395935ae79763dc349bc0eb706319d28f7a33ad2cfb3cfc178", size = 10576719, upload-time = "2026-07-16T15:13:32.35Z" },- { url = "https://files.pythonhosted.org/packages/bf/1e/3a13abd392a3b50b62e5938a831f9ab6e588358cacad5c18545b716d2182/ruff-0.15.22-py3-none-manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:62d425005c1835eb24e2ee4161cb90e8db263415f4a71c8c72c33abaa6c0c224", size = 11376494, upload-time = "2026-07-16T15:13:35.958Z" },- { url = "https://files.pythonhosted.org/packages/bf/3e/422d3d95bcf04dd78e1aeac22184d4f9a8fb2c01865d39d44618484a0317/ruff-0.15.22-py3-none-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:e8b9b3f8779a4f08c969defc3c8c35abffaa757e601ed5ae66d6d1db6519969a", size = 12208370, upload-time = "2026-07-16T15:13:39.185Z" },- { url = "https://files.pythonhosted.org/packages/1e/91/5d065a0e0a02bf4813f5119ad278462eed081d2b832eb7c021ade0ec9e65/ruff-0.15.22-py3-none-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:1e0dd1b2e4d3d585f897a0d137cbf4eaf6223bef4e8ce34d6bb12556c5f9249e", size = 11581098, upload-time = "2026-07-16T15:13:42.132Z" },- { url = "https://files.pythonhosted.org/packages/f6/f9/a0d4871d12fae702eb1f41b686caf05f1f8b124dc6db6f784f53d74918fa/ruff-0.15.22-py3-none-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:365523eb91d9224e1bcb03b022fbf0facb8f9e23792a2c53d9d4b3924bdbdebb", size = 11399422, upload-time = "2026-07-16T15:13:45.2Z" },- { url = "https://files.pythonhosted.org/packages/18/80/c843a5176cddbceb0b7e8dd41cf9993490796c1c469348d384f5a5c13c56/ruff-0.15.22-py3-none-manylinux_2_31_riscv64.whl", hash = "sha256:fabfd168afdf29fee5be98b831efa9683c94d7c5a3b58b9ce5a2e38444589a74", size = 11381683, upload-time = "2026-07-16T15:13:48.46Z" },- { url = "https://files.pythonhosted.org/packages/d4/00/8485de0ae92239438a36cfc51350db9b9e85c9ebdfaea91b18e422706662/ruff-0.15.22-py3-none-musllinux_1_2_aarch64.whl", hash = "sha256:225dbf095a87f1d9f90f5fd7924d2613ee452a75a4308c63a8f50f761787aa7c", size = 10850295, upload-time = "2026-07-16T15:13:51.655Z" },- { url = "https://files.pythonhosted.org/packages/fa/91/24977ec2ec72eaf15e4394ace2959fdff2dd1e14f03e005e838023407169/ruff-0.15.22-py3-none-musllinux_1_2_armv7l.whl", hash = "sha256:1877d63b9d24ed278744f1523fd11b85540566d54641f97c566d7d9dc5ca5296", size = 10579640, upload-time = "2026-07-16T15:13:54.79Z" },- { url = "https://files.pythonhosted.org/packages/9c/47/9b51216951974df1f263ac19da550d34252e0ed7218c25f10c5ef9ed7517/ruff-0.15.22-py3-none-musllinux_1_2_i686.whl", hash = "sha256:a1606c510bd7215680d32efab38965f7cdec3ef69f5170a3f4791404ffdd5262", size = 11105077, upload-time = "2026-07-16T15:13:57.915Z" },- { url = "https://files.pythonhosted.org/packages/c2/47/20e9d4a3b8016778acea5fc32bb50d35d207500a17ddb529ffa6996feef8/ruff-0.15.22-py3-none-musllinux_1_2_x86_64.whl", hash = "sha256:630479b18625f5ffc373f77603a22a9f8ac0acd7ff0501178b5db28ec71e9c64", size = 11490980, upload-time = "2026-07-16T15:14:01.032Z" },- { url = "https://files.pythonhosted.org/packages/4d/76/3f72d8fc38c1cb77b38c56a70da9d0c17700cc1cc50f9649c9d3c8f5ba71/ruff-0.15.22-py3-none-win32.whl", hash = "sha256:e5ba0e4a13fd14abbed2a77b517a3911290c6c6c59ef67784328d1668fab76cf", size = 10789165, upload-time = "2026-07-16T15:14:04.16Z" },- { url = "https://files.pythonhosted.org/packages/cb/46/4965251734c2b6fcdca1b1b187d20bcac3af0ee5b083b89c910bb961ce3a/ruff-0.15.22-py3-none-win_amd64.whl", hash = "sha256:9be63ba1eb936acd2d1342fb8337c356353706fce233b2a15a09a97037e6acde", size = 11938297, upload-time = "2026-07-16T15:14:07.316Z" },- { url = "https://files.pythonhosted.org/packages/57/c9/e69b1ff4c8b69093ef08b8919ab767af0569666865b39c30a8795d88d3c6/ruff-0.15.22-py3-none-win_arm64.whl", hash = "sha256:e1168075b72158510839f250027659cdd78476f40507dd517892304c41318661", size = 11298172, upload-time = "2026-07-16T15:14:10.51Z" },+version = "0.16.2"+source = { registry = "https://pypi.org/simple/" }+sdist = { url = "https://files.pythonhosted.org/packages/73/e1/4508a569211b35599016e84ba65c1a992b7a4004b4b6c4bea02a851cba1b/ruff-0.16.2.tar.gz", hash = "sha256:c3d7828d12e8927a6fc65fe38e2c2541b9e762d360a1786d752cb1b8883b3c9c", size = 4885811, upload-time = "2026-08-07T13:31:01.432Z" }+wheels = [+ { url = "https://files.pythonhosted.org/packages/14/57/db19951540f98859c956b50bdb4d31089b4d91e9f15e2968e7d5193806d5/ruff-0.16.2-py3-none-linux_armv6l.whl", hash = "sha256:3c8de4cf2181f01d57946d87d777aa52916976fc09942aed89938fab5e013318", size = 10847925, upload-time = "2026-08-07T13:30:14.468Z" },+ { url = "https://files.pythonhosted.org/packages/13/5a/995fe85a8470d3e391ac0f7fa8054bb454eaf33ee138196d6172ed1079c0/ruff-0.16.2-py3-none-macosx_10_12_x86_64.whl", hash = "sha256:9a48cc05c6fbc811ca81b5d7ba95375affea6582d1b8024e455e41afbbf55344", size = 11072662, upload-time = "2026-08-07T13:30:18.143Z" },+ { url = "https://files.pythonhosted.org/packages/32/53/370d767c61c71a971a4ace36703a7ecd8c393956349a7325d7fab2b56827/ruff-0.16.2-py3-none-macosx_11_0_arm64.whl", hash = "sha256:a2c0d14fcbb26c91f0f867a6dc9bd71bbc30b1b6151829c884f23faeab2e5700", size = 10566771, upload-time = "2026-08-07T13:30:20.899Z" },+ { url = "https://files.pythonhosted.org/packages/85/d6/9d96948caf5a632be62d62202d5ec914d6856f204fd79eb036e5915e79ea/ruff-0.16.2-py3-none-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:335c621622c4650330be50842561c6586ac6971bb8ab5407fe34dcc9efb16bbe", size = 10975825, upload-time = "2026-08-07T13:30:23.517Z" },+ { url = "https://files.pythonhosted.org/packages/3b/92/ea87129b3414acb0b5770563779c51804d37ac67675c7ba35447ddb14773/ruff-0.16.2-py3-none-manylinux_2_17_armv7l.manylinux2014_armv7l.whl", hash = "sha256:20e66910f2c37cc753f9ef6580c914a621b80c4fa3549d3e3521e29d0f5bfc3f", size = 10649437, upload-time = "2026-08-07T13:30:26.097Z" },+ { url = "https://files.pythonhosted.org/packages/ac/43/f8f291dcd4af5bb7872b74fdfa41a7cd7c856ca1d4069670971cf1b9f5cb/ruff-0.16.2-py3-none-manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:c7e36fbfba65510548156902bcf1350a979a958ce0347ce0f90d73894036b39f", size = 11446761, upload-time = "2026-08-07T13:30:28.752Z" },+ { url = "https://files.pythonhosted.org/packages/71/4a/ef991fb2fcf516ab71f0808adcdd8da5e18c8cde447f4ceaf5f47a5132a5/ruff-0.16.2-py3-none-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:f0eab35f80df8f134aae5d1630e751901321d317cc8e50dc39e36fa3ed34cd12", size = 12336364, upload-time = "2026-08-07T13:30:31.468Z" },+ { url = "https://files.pythonhosted.org/packages/f3/24/f615e74f307e6ca0e56a482872477b856c70d530aa356abfb6dfe5ca8a80/ruff-0.16.2-py3-none-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:40ea8c0594feb894e89c8c61ab9c103d38b0ea72dfde6c594107147ca31b1140", size = 11630720, upload-time = "2026-08-07T13:30:34.426Z" },+ { url = "https://files.pythonhosted.org/packages/c5/d3/8ef50149e8412a77f7ab409efdef0e2b23803707a3863da4fc64cb23d459/ruff-0.16.2-py3-none-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:ab3d62dde0b19facdd632008cc4827fc28ada7736c6bd35ab6f1050f0bfed53f", size = 11466130, upload-time = "2026-08-07T13:30:36.958Z" },+ { url = "https://files.pythonhosted.org/packages/dd/a7/a19334985c4dea8c381981fa252cd854c7ee52dc4b1686dc16f4a911c702/ruff-0.16.2-py3-none-manylinux_2_31_riscv64.whl", hash = "sha256:e43e1f5b8388da9eca1b9e88328d47a5cec794633ccf6f7484ac2dd15eee92c0", size = 11523634, upload-time = "2026-08-07T13:30:39.822Z" },+ { url = "https://files.pythonhosted.org/packages/6e/6c/96d192b0e742412ceda08c0a50f9669b253dde9fd6a60ea1a10c9fa79a63/ruff-0.16.2-py3-none-musllinux_1_2_aarch64.whl", hash = "sha256:c24788a980581e1d7ea3a0cbe4344c4fbeb0a6a9b1f4713aa46bb104f8294690", size = 10949807, upload-time = "2026-08-07T13:30:42.745Z" },+ { url = "https://files.pythonhosted.org/packages/fa/51/e26599ceca11e79ee255c7df515995561edf87e9ca1893284e44d98f5a86/ruff-0.16.2-py3-none-musllinux_1_2_armv7l.whl", hash = "sha256:81806b08329130005dd4a8a8394a0c9da8c6f4cafb16ba438d2a2ee6a18bedf1", size = 10646891, upload-time = "2026-08-07T13:30:45.522Z" },+ { url = "https://files.pythonhosted.org/packages/68/01/800c4b1f97bc8d7c6029e06b1f20473a3cf1e13c4933d8f3342add83fc55/ruff-0.16.2-py3-none-musllinux_1_2_i686.whl", hash = "sha256:4ce4e02bad779bef557f541a1b31f20d6abeae1cc05ed1b1ac019d4ffd1044c8", size = 11162063, upload-time = "2026-08-07T13:30:48.131Z" },+ { url = "https://files.pythonhosted.org/packages/e4/d0/1477ea50fc5a0d4b0b71d1d63d50770bdd794d90b43e37a7618e63ec9894/ruff-0.16.2-py3-none-musllinux_1_2_x86_64.whl", hash = "sha256:e0422abdf70070255fc4073ce9dfc814cc03db577013761ddd09bc1e4a9a4fbd", size = 11556038, upload-time = "2026-08-07T13:30:50.686Z" },+ { url = "https://files.pythonhosted.org/packages/b8/76/a7776f32048d991e16d4fa8ff91790b877342d3596cc3ed04acdbf1aaedc/ruff-0.16.2-py3-none-win32.whl", hash = "sha256:bf3a63d78fb39f4bf5ac8ae52051c5520505301abe19ba4e204c453b3f09bb0b", size = 10872850, upload-time = "2026-08-07T13:30:53.471Z" },+ { url = "https://files.pythonhosted.org/packages/00/0d/929c800d920e61397d82a01b60bffc68da3052c17d31de59efaad2e4ed75/ruff-0.16.2-py3-none-win_amd64.whl", hash = "sha256:bcabe2f6d0fc7819f1431793005af4e4de7371927d037345bf941252b195b9fa", size = 12023338, upload-time = "2026-08-07T13:30:56.193Z" },+ { url = "https://files.pythonhosted.org/packages/5b/6c/93e26c22c5f78ff87363e07da49c84955affbeb1098bd1936bf3b3f293bf/ruff-0.16.2-py3-none-win_arm64.whl", hash = "sha256:d614e95cedf38a2053fd351c55b103ba30d017d61688fdbfd40ee0412852a99f", size = 11374065, upload-time = "2026-08-07T13:30:58.775Z" },][[package]]@@ -5821,10 +5821,10 @@ resolution-markers = ["python_full_version < '3.11'",]dependencies = [- { name = "joblib" },- { name = "numpy", version = "2.2.6", source = { registry = "https://pypi.org/simple/" } },- { name = "scipy", version = "1.15.3", source = { registry = "https://pypi.org/simple/" } },- { name = "threadpoolctl" },+ { name = "joblib", marker = "python_full_version < '3.11'" },+ { name = "numpy", version = "2.2.6", source = { registry = "https://pypi.org/simple/" }, marker = "python_full_version < '3.11'" },+ { name = "scipy", version = "1.15.3", source = { registry = "https://pypi.org/simple/" }, marker = "python_full_version < '3.11'" },+ { name = "threadpoolctl", marker = "python_full_version < '3.11'" },]sdist = { url = "https://files.pythonhosted.org/packages/98/c2/a7855e41c9d285dfe86dc50b250978105dce513d6e459ea66a6aeb0e1e0c/scikit_learn-1.7.2.tar.gz", hash = "sha256:20e9e49ecd130598f1ca38a1d85090e1a600147b9c02fa6f15d69cb53d968fda", size = 7193136, upload-time = "2025-09-09T08:21:29.075Z" }wheels = [@@ -5879,10 +5879,10 @@ resolution-markers = ["python_full_version == '3.11.*' and sys_platform != 'emscripten' and sys_platform != 'win32'",]dependencies = [- { name = "joblib" },- { name = "numpy", version = "2.4.1", source = { registry = "https://pypi.org/simple/" } },- { name = "scipy", version = "1.17.0", source = { registry = "https://pypi.org/simple/" } },- { name = "threadpoolctl" },+ { name = "joblib", marker = "python_full_version >= '3.11'" },+ { name = "numpy", version = "2.4.1", source = { registry = "https://pypi.org/simple/" }, marker = "python_full_version >= '3.11'" },+ { name = "scipy", version = "1.17.0", source = { registry = "https://pypi.org/simple/" }, marker = "python_full_version >= '3.11'" },+ { name = "threadpoolctl", marker = "python_full_version >= '3.11'" },]sdist = { url = "https://files.pythonhosted.org/packages/0e/d4/40988bf3b8e34feec1d0e6a051446b1f66225f8529b9309becaeef62b6c4/scikit_learn-1.8.0.tar.gz", hash = "sha256:9bccbb3b40e3de10351f8f5068e105d0f4083b1a65fa07b6634fbc401a6287fd", size = 7335585, upload-time = "2025-12-10T07:08:53.618Z" }wheels = [@@ -5932,7 +5932,7 @@ resolution-markers = ["python_full_version < '3.11'",]dependencies = [- { name = "numpy", version = "2.2.6", source = { registry = "https://pypi.org/simple/" } },+ { name = "numpy", version = "2.2.6", source = { registry = "https://pypi.org/simple/" }, marker = "python_full_version < '3.11'" },]sdist = { url = "https://files.pythonhosted.org/packages/0f/37/6964b830433e654ec7485e45a00fc9a27cf868d622838f6b6d9c5ec0d532/scipy-1.15.3.tar.gz", hash = "sha256:eae3cf522bc7df64b42cad3925c876e1b0b6c35c1337c93e12c0f366f55b0eaf", size = 59419214, upload-time = "2025-05-08T16:13:05.955Z" }wheels = [@@ -6002,7 +6002,7 @@ resolution-markers = ["python_full_version == '3.11.*' and sys_platform != 'emscripten' and sys_platform != 'win32'",]dependencies = [- { name = "numpy", version = "2.4.1", source = { registry = "https://pypi.org/simple/" } },+ { name = "numpy", version = "2.4.1", source = { registry = "https://pypi.org/simple/" }, marker = "python_full_version >= '3.11'" },]sdist = { url = "https://files.pythonhosted.org/packages/56/3e/9cca699f3486ce6bc12ff46dc2031f1ec8eb9ccc9a320fdaf925f1417426/scipy-1.17.0.tar.gz", hash = "sha256:2591060c8e648d8b96439e111ac41fd8342fdeff1876be2e19dea3fe8930454e", size = 30396830, upload-time = "2026-01-10T21:34:23.009Z" }wheels = [diff --git a/wiki/ARCHITECTURE.md b/wiki/ARCHITECTURE.mdindex b5cd360cd..c8dccb921 100644--- a/wiki/ARCHITECTURE.md+++ b/wiki/ARCHITECTURE.md@@ -226,8 +226,8 @@ analysis = {"data": [{"ts": 45, "cpu": 92}, # Keep the spike!{"ts": 46, "cpu": 95},- ...- ]+ ...,+ ],}```@@ -397,11 +397,9 @@ def analyze_field(key, items):"unique_ratio": len(set(values)) / len(values),# 0.0 = all same (constant)# 1.0 = all different (unique IDs)-"variance": statistics.variance(values), # For numbers# Low = stable# High = changing-"change_points": detect_spikes(values),# Indices where value jumps significantly}@@ -502,14 +500,14 @@ When SmartCrusher compresses, the original content is stored for on-demand retri```python@dataclassclass CompressionEntry:- hash: str # 16-char SHA256 for retrieval- original_content: str # Full JSON before compression- compressed_content: str # Compressed JSON+ hash: str # 16-char SHA256 for retrieval+ original_content: str # Full JSON before compression+ compressed_content: str # Compressed JSONoriginal_item_count: intcompressed_item_count: int- tool_name: str | None # For feedback tracking+ tool_name: str | None # For feedback trackingcreated_at: float- ttl: int = 300 # 5 minute default+ ttl: int = 300 # 5 minute default```**Features:**@@ -625,12 +623,12 @@ The feedback system learns from retrieval patterns to improve future compression@dataclassclass ToolPattern:tool_name: str- total_compressions: int # Times we compressed this tool- total_retrievals: int # Times LLM asked for more- full_retrievals: int # Retrieved everything (all retrievals — hash-only)- search_retrievals: int # Legacy; always 0 (retrieval is hash-only, no search)- common_queries: dict[str, int] # Legacy query-pattern frequency (no longer populated)- queried_fields: dict[str, int] # Legacy queried-field frequency (no longer populated)+ total_compressions: int # Times we compressed this tool+ total_retrievals: int # Times LLM asked for more+ full_retrievals: int # Retrieved everything (all retrievals — hash-only)+ search_retrievals: int # Legacy; always 0 (retrieval is hash-only, no search)+ common_queries: dict[str, int] # Legacy query-pattern frequency (no longer populated)+ queried_fields: dict[str, int] # Legacy queried-field frequency (no longer populated)```**Key Metrics:**@@ -644,12 +642,12 @@ class ToolPattern:```python@dataclassclass CompressionHints:- max_items: int = 15 # Target item count+ max_items: int = 15 # Target item countsuggested_items: int | None # Calculated optimal- skip_compression: bool # Don't compress at all- preserve_fields: list[str] # Always keep these fields- aggressiveness: float # 0.0 = aggressive, 1.0 = conservative- reason: str # Explanation+ skip_compression: bool # Don't compress at all+ preserve_fields: list[str] # Always keep these fields+ aggressiveness: float # 0.0 = aggressive, 1.0 = conservative+ reason: str # Explanation```**Feedback-Driven Adjustment:**@@ -733,24 +731,26 @@ if self.config.use_feedback_hints and tool_name:```python@dataclassclass CCRToolCall:- tool_call_id: str # For matching response- hash_key: str # CCR hash to retrieve+ tool_call_id: str # For matching response+ hash_key: str # CCR hash to retrieve+@dataclassclass CCRToolResult:tool_call_id: str- content: str # Retrieved data as JSON+ content: str # Retrieved data as JSONsuccess: boolitems_retrieved: int+class CCRResponseHandler:async def handle_response(self,- response: dict, # Initial LLM response- messages: list, # Conversation history- tools: list, # Tool definitions- api_call_fn: Callable, # Function to make API calls- provider: str, # "anthropic" or "openai"+ response: dict, # Initial LLM response+ messages: list, # Conversation history+ tools: list, # Tool definitions+ api_call_fn: Callable, # Function to make API calls+ provider: str, # "anthropic" or "openai") -> dict:"""Handle CCR tool calls until final response."""```@@ -762,11 +762,14 @@ The handler also supports streaming responses via `StreamingCCRHandler`:```pythonclass StreamingCCRBuffer:"""Buffers streaming chunks to detect CCR tool calls."""+chunks: list[bytes]detected_ccr: bool+class StreamingCCRHandler:"""Handles CCR in streaming responses."""+async def process_stream(self, stream, messages, tools, api_call_fn):"""Yields chunks, switching to buffered mode if CCR detected."""```@@ -848,11 +851,11 @@ The tracker uses simple but effective heuristics:@dataclassclass ContextTrackerConfig:enabled: bool = True- max_tracked_contexts: int = 100 # LRU eviction- relevance_threshold: float = 0.3 # Min score to recommend- max_context_age_seconds: float = 300 # 5 minutes+ max_tracked_contexts: int = 100 # LRU eviction+ relevance_threshold: float = 0.3 # Min score to recommend+ max_context_age_seconds: float = 300 # 5 minutesproactive_expansion: bool = True- max_proactive_expansions: int = 2 # Per query+ max_proactive_expansions: int = 2 # Per query```---diff --git a/wiki/agno.md b/wiki/agno.mdindex 386d30bfe..028f36ef7 100644--- a/wiki/agno.md+++ b/wiki/agno.md@@ -167,6 +167,7 @@ Full async support for high-throughput applications:import asynciofrom headroom.integrations.agno import HeadroomAgnoModel+async def process_async():model = HeadroomAgnoModel(OpenAIChat(id="gpt-4o"))@@ -179,6 +180,7 @@ async def process_async():print(f"\nTokens saved: {model.total_tokens_saved}")+asyncio.run(process_async())```diff --git a/wiki/api.md b/wiki/api.mdindex 66288cea7..28dd29ec1 100644--- a/wiki/api.md+++ b/wiki/api.md@@ -101,9 +101,9 @@ config = CacheAlignerConfig(from headroom import RelevanceScorerConfigconfig = RelevanceScorerConfig(- scorer_type="bm25", # "bm25", "embedding", or "hybrid"- embedding_model=None, # Model name for embedding scorer- hybrid_alpha=0.5, # Weight for hybrid scoring+ scorer_type="bm25", # "bm25", "embedding", or "hybrid"+ embedding_model=None, # Model name for embedding scorer+ hybrid_alpha=0.5, # Weight for hybrid scoring)```@@ -286,10 +286,12 @@ result = aligner.align(messages)```pythonfrom headroom import TransformPipeline-pipeline = TransformPipeline([- SmartCrusher(),- CacheAligner(),-])+pipeline = TransformPipeline(+ [+ SmartCrusher(),+ CacheAligner(),+ ]+)result = pipeline.transform(messages)```diff --git a/wiki/compression.md b/wiki/compression.mdindex a76a173c3..5de8bd55e 100644--- a/wiki/compression.md+++ b/wiki/compression.md@@ -79,21 +79,18 @@ from headroom.compression import UniversalCompressorConfigconfig = UniversalCompressorConfig(# Detection- use_magika=True, # Use ML-based detection (requires magika)-+ use_magika=True, # Use ML-based detection (requires magika)# Compression# (Note: the legacy `use_llmlingua` flag was retired with the# LLMLingua-2 integration. The optional ML compressor is now Kompress,# installed via `headroom-ai[ml]` and configured separately.)compression_ratio_target=0.3, # Keep 30% of content (70% reduction)- min_content_length=100, # Skip content shorter than this-+ min_content_length=100, # Skip content shorter than this# Structure preservation- use_entropy_preservation=True, # Preserve high-entropy tokens- entropy_threshold=0.85, # Entropy threshold for preservation-+ use_entropy_preservation=True, # Preserve high-entropy tokens+ entropy_threshold=0.85, # Entropy threshold for preservation# CCR- ccr_enabled=True, # Store originals for retrieval+ ccr_enabled=True, # Store originals for retrieval)```@@ -121,12 +118,12 @@ Preserves JSON structure while compressing values:from headroom.compression.handlers.json_handler import JSONStructureHandlerhandler = JSONStructureHandler(- preserve_short_values=True, # Keep values < 20 chars- short_value_threshold=20, # Threshold for "short"- preserve_high_entropy=True, # Keep UUIDs, hashes- entropy_threshold=0.85, # Entropy threshold- max_array_items_full=3, # Keep first N array items full- max_number_digits=10, # Preserve numbers up to N digits+ preserve_short_values=True, # Keep values < 20 chars+ short_value_threshold=20, # Threshold for "short"+ preserve_high_entropy=True, # Keep UUIDs, hashes+ entropy_threshold=0.85, # Entropy threshold+ max_array_items_full=3, # Keep first N array items full+ max_number_digits=10, # Preserve numbers up to N digits)```@@ -141,18 +138,10 @@ handler = JSONStructureHandler(```python# Before-{- "id": "usr_abc123",- "name": "Alice Johnson",- "bio": "A long description that goes on and on..."-}+{"id": "usr_abc123", "name": "Alice Johnson", "bio": "A long description that goes on and on..."}# After (structure preserved, long values compressed)-{- "id": "usr_abc123",- "name": "Alice Johnson",- "bio": "A long...[compressed]..."-}+{"id": "usr_abc123", "name": "Alice Johnson", "bio": "A long...[compressed]..."}```### Code Handler@@ -163,9 +152,9 @@ Preserves code structure using AST parsing (tree-sitter) or regex fallback:from headroom.compression.handlers.code_handler import CodeStructureHandlerhandler = CodeStructureHandler(- preserve_comments=False, # Preserve comments as structural- use_tree_sitter=True, # Use tree-sitter for parsing- default_language="python", # Default when detection fails+ preserve_comments=False, # Preserve comments as structural+ use_tree_sitter=True, # Use tree-sitter for parsing+ default_language="python", # Default when detection fails)```@@ -226,24 +215,24 @@ from headroom.compression import compressresult = compress(content)# Access result fields-print(result.compressed) # Compressed content-print(result.original) # Original content-print(result.compression_ratio) # e.g., 0.35 (35% of original size)-print(result.tokens_before) # Estimated tokens before-print(result.tokens_after) # Estimated tokens after-print(result.tokens_saved) # tokens_before - tokens_after-print(result.savings_percentage) # e.g., 65.0 (65% savings)+print(result.compressed) # Compressed content+print(result.original) # Original content+print(result.compression_ratio) # e.g., 0.35 (35% of original size)+print(result.tokens_before) # Estimated tokens before+print(result.tokens_after) # Estimated tokens after+print(result.tokens_saved) # tokens_before - tokens_after+print(result.savings_percentage) # e.g., 65.0 (65% savings)# Detection info-print(result.content_type) # ContentType.JSON, CODE, etc.-print(result.detection_confidence) # 0.0-1.0+print(result.content_type) # ContentType.JSON, CODE, etc.+print(result.detection_confidence) # 0.0-1.0# Structure info-print(result.handler_used) # "json", "code", etc.-print(result.preservation_ratio) # Fraction preserved as structure+print(result.handler_used) # "json", "code", etc.+print(result.preservation_ratio) # Fraction preserved as structure# CCR info-print(result.ccr_key) # Key for retrieval (if CCR enabled)+print(result.ccr_key) # Key for retrieval (if CCR enabled)```---@@ -259,8 +248,8 @@ compressor = UniversalCompressor()contents = ['{"users": [...]}',- 'def hello(): pass',- 'Plain text content',+ "def hello(): pass",+ "Plain text content",]results = compressor.compress_batch(contents)@@ -393,11 +382,11 @@ json_content = """result = compressor.compress(json_content)-print(f"Type: {result.content_type}") # ContentType.JSON-print(f"Handler: {result.handler_used}") # json+print(f"Type: {result.content_type}") # ContentType.JSON+print(f"Handler: {result.handler_used}") # jsonprint(f"Saved: {result.savings_percentage:.0f}%") # ~60%print(f"Structure: {result.preservation_ratio:.0%} preserved") # ~40%-print(f"CCR Key: {result.ccr_key}") # For retrieval+print(f"CCR Key: {result.ccr_key}") # For retrieval```---diff --git a/wiki/configuration.md b/wiki/configuration.mdindex 58b011e04..b74c2e89f 100644--- a/wiki/configuration.md+++ b/wiki/configuration.md@@ -31,22 +31,17 @@ from openai import OpenAIclient = HeadroomClient(original_client=OpenAI(),provider=OpenAIProvider(),-# Mode: "audit" (observe only) or "optimize" (apply transforms)default_mode="optimize",-# Enable provider-specific cache optimizationenable_cache_optimizer=True,-# Enable query-level semantic cachingenable_semantic_cache=False,-# Override default context limits per modelmodel_context_limits={"gpt-4o": 128000,"gpt-4o-mini": 128000,},-# Database location (defaults to temp directory)# store_url="sqlite:////absolute/path/to/headroom.db",)@@ -136,20 +131,14 @@ Override configuration for specific requests:response = client.chat.completions.create(model="gpt-4o",messages=[...],-# Override mode for this requestheadroom_mode="audit",-# Reserve more tokens for outputheadroom_output_buffer_tokens=8000,-# Keep last N turns (don't compress)headroom_keep_turns=5,-# Skip compression for specific tools- headroom_tool_profiles={- "important_tool": {"skip_compression": True}- }+ headroom_tool_profiles={"important_tool": {"skip_compression": True}},)```@@ -186,13 +175,10 @@ from headroom.transforms import SmartCrusherConfigconfig = SmartCrusherConfig(# Maximum items to keep after compressionmax_items_after_crush=15,-# Minimum tokens before applying compressionmin_tokens_to_crush=200,-# Relevance scoring tier: "bm25" (fast) or "embedding" (accurate)relevance_tier="bm25",-# Always keep items with these field valuespreserve_fields=["error", "warning", "failure"],)@@ -208,7 +194,6 @@ from headroom.transforms import CacheAlignerConfigconfig = CacheAlignerConfig(# Enable/disable cache alignmentenabled=True,-# Patterns to extract from system promptdynamic_patterns=[r"Today is \w+ \d+, \d{4}",diff --git a/wiki/errors.md b/wiki/errors.mdindex 345e593a5..c5642d362 100644--- a/wiki/errors.md+++ b/wiki/errors.md@@ -6,12 +6,12 @@ Headroom provides explicit exceptions for debugging, with a safety guarantee tha```pythonfrom headroom import (- HeadroomError, # Base class - catch all Headroom errors- ConfigurationError, # Invalid configuration- ProviderError, # Provider issues (unknown model, etc.)- StorageError, # Database/storage failures- CompressionError, # Compression failures (rare)- ValidationError, # Setup validation failures+ HeadroomError, # Base class - catch all Headroom errors+ ConfigurationError, # Invalid configuration+ ProviderError, # Provider issues (unknown model, etc.)+ StorageError, # Database/storage failures+ CompressionError, # Compression failures (rare)+ ValidationError, # Setup validation failures)```@@ -75,10 +75,7 @@ Raised for provider-specific issues.# - Token counting failuretry:- response = client.chat.completions.create(- model="unknown-model-xyz",- messages=[...]- )+ response = client.chat.completions.create(model="unknown-model-xyz", messages=[...])except ProviderError as e:print(f"Provider error: {e}")print(f"Provider: {e.details.get('provider')}")@@ -122,10 +119,7 @@ Raised when setup validation fails.```pythonresult = client.validate_setup()if not result["valid"]:- raise ValidationError(- "Setup validation failed",- details={"issues": result["issues"]}- )+ raise ValidationError("Setup validation failed", details={"issues": result["issues"]})```## Safety Guarantee@@ -136,16 +130,11 @@ This is a core design principle. Your LLM calls never fail due to Headroom:```python# Even if SmartCrusher encounters unexpected data:-messages = [- {"role": "tool", "content": "malformed json {{{"}-]+messages = [{"role": "tool", "content": "malformed json {{{"}]# This will NOT raise an exception# Instead, the malformed content passes through unchanged-response = client.chat.completions.create(- model="gpt-4o",- messages=messages-)+response = client.chat.completions.create(model="gpt-4o", messages=messages)```## Logging Errors@@ -154,6 +143,7 @@ Enable logging to see error details:```pythonimport logging+logging.basicConfig(level=logging.WARNING)# Now you'll see warnings when compression is skipped:@@ -232,6 +222,7 @@ response = client.chat.completions.create(...)```pythonimport logging+logging.basicConfig(level=logging.DEBUG)# Shows detailed transform decisionsdiff --git a/wiki/image-compression.md b/wiki/image-compression.mdindex 4d50b8b30..0fa22ae9c 100644--- a/wiki/image-compression.md+++ b/wiki/image-compression.md@@ -61,13 +61,15 @@ client = HeadroomClient(provider="openai")response = client.chat.completions.create(model="gpt-4o",- messages=[{- "role": "user",- "content": [- {"type": "text", "text": "What animal is this?"},- {"type": "image_url", "image_url": {"url": "data:image/jpeg;base64,..."}}- ]- }]+ messages=[+ {+ "role": "user",+ "content": [+ {"type": "text", "text": "What animal is this?"},+ {"type": "image_url", "image_url": {"url": "data:image/jpeg;base64,..."}},+ ],+ }+ ],)# Image automatically compressed with detail="low" (87% savings)```@@ -106,8 +108,8 @@ from headroom.image import ImageCompressorcompressor = ImageCompressor(model_id="chopratejas/technique-router", # HuggingFace model- use_siglip=True, # Enable image analysis- device="cuda", # Use GPU if available+ use_siglip=True, # Enable image analysis+ device="cuda", # Use GPU if available)```@@ -229,9 +231,11 @@ The HuggingFace model downloads on first use:```python# Force a specific cache directoryimport os+os.environ["HF_HOME"] = "/path/to/cache"from headroom.image import ImageCompressor+compressor = ImageCompressor()```@@ -290,10 +294,10 @@ class ImageCompressor:```python@dataclassclass CompressionResult:- technique: Technique # full_low, preserve, crop, transcode- original_tokens: int # Estimated tokens before- compressed_tokens: int # Estimated tokens after- confidence: float # Router confidence (0-1)+ technique: Technique # full_low, preserve, crop, transcode+ original_tokens: int # Estimated tokens before+ compressed_tokens: int # Estimated tokens after+ confidence: float # Router confidence (0-1)@propertydef savings_percent(self) -> float:@@ -304,10 +308,10 @@ class CompressionResult:```pythonclass Technique(Enum):- FULL_LOW = "full_low" # 87% savings- PRESERVE = "preserve" # 0% savings- CROP = "crop" # 50-90% savings- TRANSCODE = "transcode" # 99% savings+ FULL_LOW = "full_low" # 87% savings+ PRESERVE = "preserve" # 0% savings+ CROP = "crop" # 50-90% savings+ TRANSCODE = "transcode" # 99% savings```## See Alsodiff --git a/wiki/integration-guide.md b/wiki/integration-guide.mdindex c1d0885bf..6d076ab06 100644--- a/wiki/integration-guide.md+++ b/wiki/integration-guide.md@@ -92,22 +92,26 @@ import httpxfrom headroom import compresscompressed = compress(messages, model="claude-sonnet-4-5-20250929")-httpx.post("https://api.anthropic.com/v1/messages", json={- "model": "claude-sonnet-4-5-20250929",- "messages": compressed.messages,-}, headers={"X-Api-Key": api_key, "anthropic-version": "2023-06-01"})+httpx.post(+ "https://api.anthropic.com/v1/messages",+ json={+ "model": "claude-sonnet-4-5-20250929",+ "messages": compressed.messages,+ },+ headers={"X-Api-Key": api_key, "anthropic-version": "2023-06-01"},+)```### What compress() returns```pythonresult = compress(messages, model="gpt-4o")-result.messages # list[dict] — compressed messages, same format as input-result.tokens_before # int — original token count-result.tokens_after # int — compressed token count-result.tokens_saved # int — tokens removed+result.messages # list[dict] — compressed messages, same format as input+result.tokens_before # int — original token count+result.tokens_after # int — compressed token count+result.tokens_saved # int — tokens removedresult.compression_ratio # float — 0.0 (no savings) to 1.0 (100% removed)-result.transforms_applied # list[str] — what ran (e.g., ["router:smart_crusher:0.35"])+result.transforms_applied # list[str] — what ran (e.g., ["router:smart_crusher:0.35"])```---@@ -169,6 +173,7 @@ app.add_middleware(CompressionMiddleware)# LiteLLM proxyfrom litellm.proxy.proxy_server import app+app.add_middleware(CompressionMiddleware)```@@ -309,6 +314,7 @@ Customize compression behavior without modifying Headroom's code:```pythonfrom headroom import compress, CompressionHooks, CompressContext+class MyHooks(CompressionHooks):def pre_compress(self, messages, ctx):# Modify messages before compression (dedup, filter, inject)@@ -323,6 +329,7 @@ class MyHooks(CompressionHooks):# Observe results (logging, analytics, learning)print(f"Saved {event.tokens_saved} tokens")+result = compress(messages, model="gpt-4o", hooks=MyHooks())```diff --git a/wiki/langchain.md b/wiki/langchain.mdindex 1d79f1da3..c89bf3fc8 100644--- a/wiki/langchain.md+++ b/wiki/langchain.md@@ -91,11 +91,13 @@ Works seamlessly with LangChain tool calling:```pythonfrom langchain_core.tools import tool+@tooldef search(query: str) -> str:"""Search the web."""return {"results": [...]} # Large JSON response+llm_with_tools = llm.bind_tools([search])response = llm_with_tools.invoke("Search for Python tutorials")# Tool outputs are automatically compressed in subsequent turns@@ -117,7 +119,7 @@ base_history = ChatMessageHistory()compressed_history = HeadroomChatMessageHistory(base_history,compress_threshold_tokens=4000, # Compress when over 4K tokens- keep_recent_turns=5, # Always keep last 5 turns+ keep_recent_turns=5, # Always keep last 5 turns)# Use with any memory class@@ -152,9 +154,9 @@ base_retriever = vectorstore.as_retriever(search_kwargs={"k": 50})# Wrap with Headroom compression (keep best for precision)compressor = HeadroomDocumentCompressor(- max_documents=10, # Keep top 10- min_relevance=0.3, # Minimum relevance score- prefer_diverse=True, # MMR-style diversity+ max_documents=10, # Keep top 10+ min_relevance=0.3, # Minimum relevance score+ prefer_diverse=True, # MMR-style diversity)retriever = ContextualCompressionRetriever(@@ -179,18 +181,21 @@ from langchain.agents import create_openai_tools_agent, AgentExecutorfrom langchain_core.tools import toolfrom headroom.integrations import wrap_tools_with_headroom+@tooldef search_database(query: str) -> str:"""Search the database."""# Returns 1000 results as JSONreturn json.dumps({"results": [...], "total": 1000})+@tooldef fetch_logs(service: str) -> str:"""Fetch service logs."""# Returns 500 log entriesreturn json.dumps({"logs": [...]})+# Wrap tools with compressiontools = [search_database, fetch_logs]wrapped_tools = wrap_tools_with_headroom(@@ -296,26 +301,33 @@ from langchain_core.tools import toolfrom langgraph.prebuilt import create_react_agentfrom headroom.integrations import HeadroomChatModel, wrap_tools_with_headroom+# Define tools that return large outputs@tooldef search_web(query: str) -> str:"""Search the web for information."""# Simulating large search results- return json.dumps({- "results": [- {"title": f"Result {i}", "snippet": "..." * 100, "url": f"https://..."}- for i in range(100)- ],- "total": 1000,- })+ return json.dumps(+ {+ "results": [+ {"title": f"Result {i}", "snippet": "..." * 100, "url": f"https://..."}+ for i in range(100)+ ],+ "total": 1000,+ }+ )+@tooldef query_database(sql: str) -> str:"""Execute SQL query."""- return json.dumps({- "rows": [{"id": i, "data": "..." * 50} for i in range(500)],- "total": 500,- })+ return json.dumps(+ {+ "rows": [{"id": i, "data": "..." * 50} for i in range(500)],+ "total": 500,+ }+ )+# Wrap model with Headroomllm = HeadroomChatModel(ChatOpenAI(model="gpt-4o"))@@ -327,9 +339,9 @@ tools = wrap_tools_with_headroom([search_web, query_database])agent = create_react_agent(llm, tools)# Run - tool outputs are automatically compressed between iterations-result = agent.invoke({- "messages": [("user", "Find all users who signed up last week and their activity")]-})+result = agent.invoke(+ {"messages": [("user", "Find all users who signed up last week and their activity")]}+)# Check savingsprint(f"Tokens saved: {llm.get_metrics()['tokens_saved']}")@@ -352,23 +364,29 @@ from langchain_core.messages import HumanMessagefrom langgraph.graph import StateGraph, MessagesState, START, ENDfrom headroom.integrations.langchain import create_compress_tool_messages_node+# Define your agent and tools nodesdef agent_node(state: MessagesState):llm = ChatOpenAI(model="gpt-4o")response = llm.invoke(state["messages"])return {"messages": [response]}+def tools_node(state: MessagesState):# Your tool execution logic here...+# Build the graph with a compression stepgraph = StateGraph(MessagesState)graph.add_node("agent", agent_node)graph.add_node("tools", tools_node)-graph.add_node("compress", create_compress_tool_messages_node(- min_tokens_to_compress=100, # Only compress outputs > ~100 tokens-))+graph.add_node(+ "compress",+ create_compress_tool_messages_node(+ min_tokens_to_compress=100, # Only compress outputs > ~100 tokens+ ),+)# Wire: tools -> compress -> agent (instead of tools -> agent directly)graph.add_edge(START, "agent")@@ -411,9 +429,9 @@ base_retriever = vectorstore.as_retriever(search_kwargs={"k": 50})# Headroom compressor for precisioncompressor = HeadroomDocumentCompressor(- max_documents=5, # Keep only top 5- min_relevance=0.4, # Must be 40%+ relevant- prefer_diverse=True, # Avoid redundant docs+ max_documents=5, # Keep only top 5+ min_relevance=0.4, # Must be 40%+ relevant+ prefer_diverse=True, # Avoid redundant docs)# Combine into compression retriever@@ -461,7 +479,7 @@ base_history = ChatMessageHistory()compressed_history = HeadroomChatMessageHistory(base_history,compress_threshold_tokens=8000, # Compress when over 8K- keep_recent_turns=10, # Always keep last 10 turns+ keep_recent_turns=10, # Always keep last 10 turns)memory = ConversationBufferMemory(@@ -500,30 +518,45 @@ from headroom.integrations import (reset_tool_metrics,)+@tooldef search_arxiv(query: str) -> str:"""Search arXiv for papers."""- return json.dumps({"papers": [{"title": f"Paper {i}", "abstract": "..." * 200} for i in range(50)]})+ return json.dumps(+ {"papers": [{"title": f"Paper {i}", "abstract": "..." * 200} for i in range(50)]}+ )+@tooldef search_github(query: str) -> str:"""Search GitHub repositories."""- return json.dumps({"repos": [{"name": f"repo-{i}", "description": "..." * 100, "stars": i * 100} for i in range(100)]})+ return json.dumps(+ {+ "repos": [+ {"name": f"repo-{i}", "description": "..." * 100, "stars": i * 100}+ for i in range(100)+ ]+ }+ )+@tooldef fetch_documentation(url: str) -> str:"""Fetch documentation from URL."""return "..." * 5000 # Large doc content+# Wrap everythingllm = HeadroomChatModel(ChatOpenAI(model="gpt-4o"))tools = wrap_tools_with_headroom([search_arxiv, search_github, fetch_documentation])-prompt = ChatPromptTemplate.from_messages([- ("system", "You are a research assistant. Use tools to gather information."),- ("human", "{input}"),- ("placeholder", "{agent_scratchpad}"),-])+prompt = ChatPromptTemplate.from_messages(+ [+ ("system", "You are a research assistant. Use tools to gather information."),+ ("human", "{input}"),+ ("placeholder", "{agent_scratchpad}"),+ ]+)agent = create_openai_tools_agent(llm, tools, prompt)executor = AgentExecutor(agent=agent, tools=tools, verbose=True)@@ -532,9 +565,11 @@ executor = AgentExecutor(agent=agent, tools=tools, verbose=True)reset_tool_metrics()# Run complex research task-result = executor.invoke({- "input": "Research the latest advances in LLM context compression and find relevant GitHub projects"-})+result = executor.invoke(+ {+ "input": "Research the latest advances in LLM context compression and find relevant GitHub projects"+ }+)# Check per-tool metricsmetrics = get_tool_metrics().get_summary()@@ -550,9 +585,9 @@ print(f"Per-tool breakdown: {metrics['by_tool']}")```pythonHeadroomChatModel(- wrapped_model, # Any LangChain BaseChatModel+ wrapped_model, # Any LangChain BaseChatModelheadroom_config=HeadroomConfig(), # Headroom configuration- auto_detect_provider=True, # Auto-detect from wrapped model+ auto_detect_provider=True, # Auto-detect from wrapped model)```@@ -560,10 +595,10 @@ HeadroomChatModel(```pythonHeadroomChatMessageHistory(- base_history, # Any BaseChatMessageHistory- compress_threshold_tokens=4000, # Token threshold for compression- keep_recent_turns=5, # Minimum turns to preserve- model="gpt-4o", # Model for token counting+ base_history, # Any BaseChatMessageHistory+ compress_threshold_tokens=4000, # Token threshold for compression+ keep_recent_turns=5, # Minimum turns to preserve+ model="gpt-4o", # Model for token counting)```@@ -571,9 +606,9 @@ HeadroomChatMessageHistory(```pythonHeadroomDocumentCompressor(- max_documents=10, # Maximum docs to return- min_relevance=0.0, # Minimum relevance score (0-1)- prefer_diverse=False, # Use MMR for diversity+ max_documents=10, # Maximum docs to return+ min_relevance=0.0, # Minimum relevance score (0-1)+ prefer_diverse=False, # Use MMR for diversity)```@@ -581,9 +616,9 @@ HeadroomDocumentCompressor(```pythonwrap_tools_with_headroom(- tools, # List of LangChain tools- min_chars_to_compress=1000, # Minimum output size- smart_crusher_config=None, # SmartCrusher configuration+ tools, # List of LangChain tools+ min_chars_to_compress=1000, # Minimum output size+ smart_crusher_config=None, # SmartCrusher configuration)```@@ -595,27 +630,21 @@ wrap_tools_with_headroom(from headroom.integrations import (# Chat ModelHeadroomChatModel,-# MemoryHeadroomChatMessageHistory,-# RetrieversHeadroomDocumentCompressor,-# AgentsHeadroomToolWrapper,wrap_tools_with_headroom,get_tool_metrics,reset_tool_metrics,-# StreamingStreamingMetricsTracker,StreamingMetricsCallback,track_streaming_response,-# LangSmithHeadroomLangSmithCallbackHandler,-# Provider Detectiondetect_provider,get_headroom_provider,@@ -660,7 +689,7 @@ Check that your message count exceeds the threshold:history = HeadroomChatMessageHistory(base_history,compress_threshold_tokens=1000, # Lower threshold- keep_recent_turns=2, # Fewer preserved turns+ keep_recent_turns=2, # Fewer preserved turns)```diff --git a/wiki/memory.md b/wiki/memory.mdindex f699c4015..30d35e7f7 100644--- a/wiki/memory.md+++ b/wiki/memory.md@@ -159,15 +159,13 @@ client = with_memory(OpenAI(), user_id="alice")# Use exactly like normalresponse = client.chat.completions.create(- model="gpt-4o",- messages=[{"role": "user", "content": "I prefer Python for backend work"}]+ model="gpt-4o", messages=[{"role": "user", "content": "I prefer Python for backend work"}])# Memory extracted INLINE - zero extra latency# Later, in a new conversation...response = client.chat.completions.create(- model="gpt-4o",- messages=[{"role": "user", "content": "What language should I use?"}]+ model="gpt-4o", messages=[{"role": "user", "content": "What language should I use?"}])# → Response uses the Python preference from memory```@@ -228,7 +226,7 @@ client1 = with_memory()response = client1.chat.completions.create(model="gpt-4o",- messages=[{"role": "user", "content": "I prefer Go for performance-critical code"}]+ messages=[{"role": "user", "content": "I prefer Go for performance-critical code"}],)# Memory stored at USER level (persists across sessions)@@ -239,8 +237,7 @@ client2 = with_memory(session_id="afternoon-session", # Different session)response = client2.chat.completions.create(- model="gpt-4o",- messages=[{"role": "user", "content": "What language for my new microservice?"}]+ model="gpt-4o", messages=[{"role": "user", "content": "What language for my new microservice?"}])# → Recalls Go preference from morning session!```@@ -270,17 +267,21 @@ new = await memory.supersede()# Query current state (excludes superseded)-current = await memory.query(MemoryFilter(- user_id="alice",- include_superseded=False, # Default-))+current = await memory.query(+ MemoryFilter(+ user_id="alice",+ include_superseded=False, # Default+ )+)# → Returns only "User now works at Anthropic"# Query full history (includes superseded)-history = await memory.query(MemoryFilter(- user_id="alice",- include_superseded=True,-))+history = await memory.query(+ MemoryFilter(+ user_id="alice",+ include_superseded=True,+ )+)# → Returns both memories with validity timestamps# Get the chain@@ -362,6 +363,7 @@ from headroom.memory import ()from headroom.memory.ports import MemoryFilter, VectorFilter+async def main():# Create with custom configurationconfig = MemoryConfig(@@ -402,17 +404,20 @@ async def main():)# Query with filters- memories = await memory.query(MemoryFilter(- user_id="alice",- categories=[MemoryCategory.PREFERENCE, MemoryCategory.FACT],- min_importance=0.7,- limit=10,- ))+ memories = await memory.query(+ MemoryFilter(+ user_id="alice",+ categories=[MemoryCategory.PREFERENCE, MemoryCategory.FACT],+ min_importance=0.7,+ limit=10,+ )+ )# Convenience methodsawait memory.remember("Likes coffee", user_id="alice", importance=0.6)relevant = await memory.recall("beverage preferences", user_id="alice")+asyncio.run(main())```@@ -482,13 +487,13 @@ Apple GPU instead of the default ONNX CPU embedder. Notes:```pythonconfig = MemoryConfig(- db_path="memory.db", # SQLite database path- vector_dimension=384, # Must match embedder output- hnsw_ef_construction=200, # HNSW index quality (higher = better, slower)- hnsw_m=16, # HNSW connections per node- hnsw_ef_search=50, # HNSW search quality- cache_enabled=True, # Enable LRU cache- cache_max_size=1000, # Max cached memories+ db_path="memory.db", # SQLite database path+ vector_dimension=384, # Must match embedder output+ hnsw_ef_construction=200, # HNSW index quality (higher = better, slower)+ hnsw_m=16, # HNSW connections per node+ hnsw_ef_search=50, # HNSW search quality+ cache_enabled=True, # Enable LRU cache+ cache_max_size=1000, # Max cached memories)```@@ -499,7 +504,7 @@ client = with_memory(OpenAI(),user_id="alice",db_path="memory.db",- top_k=5, # Memories to inject per request+ top_k=5, # Memories to inject per requestsession_id="optional-session",agent_id="optional-agent",embedder_backend=EmbedderBackend.LOCAL,@@ -659,6 +664,7 @@ client = with_memory(# Groqfrom groq import Groq+client = with_memory(Groq(), user_id="alice")# Any OpenAI-compatible client@@ -678,10 +684,12 @@ client = with_memory(OpenAI(), user_id="developer_jane")# Conversation 1: User shares contextresponse = client.chat.completions.create(model="gpt-4o",- messages=[{- "role": "user",- "content": "I'm a Python developer at a fintech startup. We use PostgreSQL and FastAPI."- }]+ messages=[+ {+ "role": "user",+ "content": "I'm a Python developer at a fintech startup. We use PostgreSQL and FastAPI.",+ }+ ],)# Memories extracted:# - [FACT] Python developer at fintech startup@@ -691,10 +699,7 @@ response = client.chat.completions.create(# Conversation 2 (new session): User asks questionresponse = client.chat.completions.create(model="gpt-4o",- messages=[{- "role": "user",- "content": "What database should I use for my new project?"- }]+ messages=[{"role": "user", "content": "What database should I use for my new project?"}],)# Response references PostgreSQL preference from memory:# → "Given your experience with PostgreSQL at your fintech company,diff --git a/wiki/metrics.md b/wiki/metrics.mdindex 01c09612e..12f8d0c5f 100644--- a/wiki/metrics.md+++ b/wiki/metrics.md@@ -336,19 +336,19 @@ print(stats)"tokens_saved_total": 15000,"tokens_output_total": 8000,"cache_hits": 3,- "compression_ratio_avg": 0.70+ "compression_ratio_avg": 0.70,},"config": {"mode": "optimize","provider": "openai","cache_optimizer_enabled": True,- "semantic_cache_enabled": False+ "semantic_cache_enabled": False,},"transforms": {"smart_crusher_enabled": True,"cache_aligner_enabled": True,- "rolling_window_enabled": True- }+ "rolling_window_enabled": True,+ },}```diff --git a/wiki/plans/dynamic-smart-crusher.md b/wiki/plans/dynamic-smart-crusher.mdindex 88754bf79..648927eb2 100644--- a/wiki/plans/dynamic-smart-crusher.md+++ b/wiki/plans/dynamic-smart-crusher.md@@ -72,17 +72,18 @@ Different data patterns need different position importance:```pythonclass AnchorStrategy(Enum):- FRONT_HEAVY = "front_heavy" # Search results: top items matter most- BACK_HEAVY = "back_heavy" # Logs: recent items matter most- BALANCED = "balanced" # Time series: both ends matter- MIDDLE_AWARE = "middle_aware" # Database: order might be arbitrary+ FRONT_HEAVY = "front_heavy" # Search results: top items matter most+ BACK_HEAVY = "back_heavy" # Logs: recent items matter most+ BALANCED = "balanced" # Time series: both ends matter+ MIDDLE_AWARE = "middle_aware" # Database: order might be arbitrary+def get_anchor_strategy(pattern: DataPattern) -> AnchorStrategy:return {DataPattern.SEARCH_RESULTS: AnchorStrategy.FRONT_HEAVY, # Top N by score- DataPattern.LOGS: AnchorStrategy.BACK_HEAVY, # Recency matters- DataPattern.TIME_SERIES: AnchorStrategy.BALANCED, # Both ends for trend- DataPattern.GENERIC: AnchorStrategy.MIDDLE_AWARE, # Don't assume order+ DataPattern.LOGS: AnchorStrategy.BACK_HEAVY, # Recency matters+ DataPattern.TIME_SERIES: AnchorStrategy.BALANCED, # Both ends for trend+ DataPattern.GENERIC: AnchorStrategy.MIDDLE_AWARE, # Don't assume order}.get(pattern, AnchorStrategy.BALANCED)```@@ -150,7 +151,7 @@ def select_informative_anchors(items: list[dict],region: str, # "front", "back", "middle"slots: int,- all_items_hash: set[str]+ all_items_hash: set[str],) -> list[int]:"""Select most informative items from a region."""@@ -195,7 +196,7 @@ def calculate_information_score(item: dict, all_items: list[dict]) -> float:for field, value in item.items():field_values = [i.get(field) for i in all_items if field in i]value_frequency = field_values.count(value) / len(field_values)- score += (1 - value_frequency) # Rare values score higher+ score += 1 - value_frequency # Rare values score higher# 2. Structural uniqueness - different fields than typicaltypical_fields = get_typical_fields(all_items)@@ -221,6 +222,7 @@ Track which positions users actually retrieve and learn from it:@dataclassclass PositionRetrievalPattern:"""Learned position importance from retrieval data."""+tool_name: strtotal_compressions: intposition_retrievals: dict[str, int] # "front_10%", "middle", "back_10%"@@ -231,21 +233,13 @@ class PositionRetrievalPattern:if total == 0:return {"front": 0.5, "middle": 0.0, "back": 0.5}- return {- position: count / total- for position, count in self.position_retrievals.items()- }+ return {position: count / total for position, count in self.position_retrievals.items()}class TOINPositionLearning:"""Learn position importance from retrieval patterns."""- def record_retrieval(- self,- tool_name: str,- original_size: int,- retrieved_indices: list[int]- ):+ def record_retrieval(self, tool_name: str, original_size: int, retrieved_indices: list[int]):"""Record which positions were retrieved."""for idx in retrieved_indices:position = self._classify_position(idx, original_size)@@ -287,9 +281,7 @@ For large arrays, sample strategically from middle:```pythondef stratified_middle_sample(- items: list[dict],- num_samples: int,- analysis: ArrayAnalysis+ items: list[dict], num_samples: int, analysis: ArrayAnalysis) -> list[int]:"""Sample middle positions using stratified approach."""@@ -310,9 +302,7 @@ def stratified_middle_sample(if analysis.numeric_fields:variance_scores = calculate_position_variance(items, analysis.numeric_fields)sorted_by_variance = sorted(- middle_items,- key=lambda i: variance_scores.get(i, 0),- reverse=True+ middle_items, key=lambda i: variance_scores.get(i, 0), reverse=True)return sorted(sorted_by_variance[:num_samples])@@ -383,11 +373,7 @@ class TestAdversarialPositions:items[42]["name"] = "target_item"items[42]["description"] = "This is what user asked about"- result = smart_crusher.crush(- items,- max_items=10,- query="find target_item"- )+ result = smart_crusher.crush(items, max_items=10, query="find target_item")# Query-matched item MUST be preservedassert any("target_item" in item.get("name", "") for item in result)@@ -399,12 +385,15 @@ class TestAdversarialPositions:class TestSizeAdaptation:"""Test that anchor allocation scales with array size."""- @pytest.mark.parametrize("size,expected_min_anchors", [- (20, 3), # Small array: at least 3 anchors- (100, 4), # Medium array: at least 4 anchors- (500, 5), # Large array: at least 5 anchors- (2000, 6), # Very large: at least 6 anchors- ])+ @pytest.mark.parametrize(+ "size,expected_min_anchors",+ [+ (20, 3), # Small array: at least 3 anchors+ (100, 4), # Medium array: at least 4 anchors+ (500, 5), # Large array: at least 5 anchors+ (2000, 6), # Very large: at least 6 anchors+ ],+ )def test_anchor_count_scales(self, size, expected_min_anchors):"""Anchor count should increase with array size."""items = [{"id": i, "value": i * 10} for i in range(size)]@@ -413,8 +402,7 @@ class TestSizeAdaptation:# Count items from first 10% and last 10%anchor_count = sum(- 1 for item in result- if item["id"] < size * 0.1 or item["id"] > size * 0.9+ 1 for item in result if item["id"] < size * 0.1 or item["id"] > size * 0.9)assert anchor_count >= expected_min_anchors@@ -454,10 +442,7 @@ class TestPatternAwareAnchoring:def test_search_results_front_heavy(self):"""Search results should preserve more from front."""- items = [- {"title": f"Result {i}", "score": 1.0 - (i * 0.01)}- for i in range(100)- ]+ items = [{"title": f"Result {i}", "score": 1.0 - (i * 0.01)} for i in range(100)]result = smart_crusher.crush(items, max_items=10)@@ -485,10 +470,7 @@ class TestPatternAwareAnchoring:def test_time_series_balanced(self):"""Time series should have balanced front/back."""- items = [- {"timestamp": f"2024-01-01T{i:02d}:00:00", "value": 100 + i}- for i in range(24)- ]+ items = [{"timestamp": f"2024-01-01T{i:02d}:00:00", "value": 100 + i} for i in range(24)]result = smart_crusher.crush(items, max_items=8)@@ -510,11 +492,7 @@ class TestQueryAwareAnchoring:"""'Latest' in query should preserve more recent items."""items = [{"id": i, "created": f"2024-01-{i:02d}"} for i in range(1, 31)]- result = smart_crusher.crush(- items,- max_items=8,- query="Show me the latest entries"- )+ result = smart_crusher.crush(items, max_items=8, query="Show me the latest entries")ids = [item["id"] for item in result]recent_count = sum(1 for id in ids if id > 20)@@ -525,11 +503,7 @@ class TestQueryAwareAnchoring:"""'First' in query should preserve earlier items."""items = [{"id": i, "created": f"2024-01-{i:02d}"} for i in range(1, 31)]- result = smart_crusher.crush(- items,- max_items=8,- query="Show me the first entries"- )+ result = smart_crusher.crush(items, max_items=8, query="Show me the first entries")ids = [item["id"] for item in result]early_count = sum(1 for id in ids if id < 10)@@ -540,11 +514,7 @@ class TestQueryAwareAnchoring:"""Query for specific ID should find it regardless of position."""items = [{"id": f"item_{i:04d}", "value": i} for i in range(1000)]- result = smart_crusher.crush(- items,- max_items=10,- query="Find item_0567"- )+ result = smart_crusher.crush(items, max_items=10, query="Find item_0567")assert any(item["id"] == "item_0567" for item in result)```@@ -574,8 +544,7 @@ class TestCoverageMetrics:def test_category_coverage(self):"""Preserved items should represent all categories."""items = [- {"category": cat, "id": i}- for i, cat in enumerate(["A"] * 30 + ["B"] * 30 + ["C"] * 40)+ {"category": cat, "id": i} for i, cat in enumerate(["A"] * 30 + ["B"] * 30 + ["C"] * 40)]result = smart_crusher.crush(items, max_items=10)@@ -608,10 +577,7 @@ class TestRetrievalSimulation:def test_retrieval_hit_rate_random_queries(self):"""Measure how often preserved items satisfy random queries."""- items = [- {"id": i, "name": f"Item {i}", "category": f"cat_{i % 5}"}- for i in range(100)- ]+ items = [{"id": i, "name": f"Item {i}", "category": f"cat_{i % 5}"} for i in range(100)]compressed = smart_crusher.crush(items, max_items=15)compressed_ids = {item["id"] for item in compressed}@@ -636,9 +602,9 @@ class TestRetrievalSimulation:# Weight queries toward front (30%), back (30%), anomalies (40%)hits = 0queries = (- list(range(10)) * 3 + # Front queries- list(range(90, 100)) * 3 + # Back queries- [50] * 4 # Middle anomaly queries+ list(range(10)) * 3 # Front queries+ + list(range(90, 100)) * 3 # Back queries+ + [50] * 4 # Middle anomaly queries)for target_id in queries:@@ -718,12 +684,12 @@ class AnchorConfig:time_series_balance: float = 0.5# Query keyword detection- recency_keywords: list[str] = field(default_factory=lambda: [- "latest", "recent", "last", "newest", "current"- ])- historical_keywords: list[str] = field(default_factory=lambda: [- "first", "oldest", "earliest", "original", "initial"- ])+ recency_keywords: list[str] = field(+ default_factory=lambda: ["latest", "recent", "last", "newest", "current"]+ )+ historical_keywords: list[str] = field(+ default_factory=lambda: ["first", "oldest", "earliest", "original", "initial"]+ )# Information density selectionuse_information_density: bool = Truediff --git a/wiki/proxy.md b/wiki/proxy.mdindex 184640916..03c586b68 100644--- a/wiki/proxy.md+++ b/wiki/proxy.md@@ -323,13 +323,18 @@ The provider caches the bytes you *forwarded*, which compression already changed```pythonforwarded = []++def next_turn(new_messages):- r = requests.post(f"{proxy}/v1/compress", json={- "messages": forwarded + new_messages,- "model": "claude-sonnet-4-6",- "config": {"frozen_message_count": len(forwarded)},- }).json()- forwarded[:] = r["messages"] # next turn's frozen prefix+ r = requests.post(+ f"{proxy}/v1/compress",+ json={+ "messages": forwarded + new_messages,+ "model": "claude-sonnet-4-6",+ "config": {"frozen_message_count": len(forwarded)},+ },+ ).json()+ forwarded[:] = r["messages"] # next turn's frozen prefixreturn forwarded```diff --git a/wiki/quickstart.md b/wiki/quickstart.mdindex 417b779b0..405b32404 100644--- a/wiki/quickstart.md+++ b/wiki/quickstart.md@@ -149,19 +149,21 @@ messages = [{"role": "assistant","content": None,- "tool_calls": [{- "id": "call_1",- "type": "function",- "function": {"name": "search", "arguments": '{"q": "python"}'},- }],+ "tool_calls": [+ {+ "id": "call_1",+ "type": "function",+ "function": {"name": "search", "arguments": '{"q": "python"}'},+ }+ ],},{"role": "tool","tool_call_id": "call_1",# This is where Headroom shines - compressing large outputs- "content": json.dumps({- "results": [{"title": f"Result {i}", "score": 100-i} for i in range(500)]- }),+ "content": json.dumps(+ {"results": [{"title": f"Result {i}", "score": 100 - i} for i in range(500)]}+ ),},{"role": "user", "content": "What are the top 3 results?"},]@@ -188,7 +190,9 @@ plan = client.chat.completions.simulate(print(f"Tokens before: {plan.tokens_before}")print(f"Tokens after: {plan.tokens_after}")-print(f"Would save: {plan.tokens_saved} tokens ({plan.tokens_saved/plan.tokens_before*100:.0f}%)")+print(+ f"Would save: {plan.tokens_saved} tokens ({plan.tokens_saved / plan.tokens_before * 100:.0f}%)"+)print(f"Transforms: {plan.transforms}")print(f"Estimated savings: {plan.estimated_savings}")```@@ -227,6 +231,7 @@ print(response.content[0].text)```pythonimport logging+logging.basicConfig(level=logging.INFO)# Now you'll see:@@ -347,6 +352,7 @@ print(stats["config"]["mode"]) # Should be "optimize"# 2. Enable logging to see what's happeningimport logging+logging.basicConfig(level=logging.DEBUG)```diff --git a/wiki/sdk.md b/wiki/sdk.mdindex 81fe5af90..674d5c3e6 100644--- a/wiki/sdk.md+++ b/wiki/sdk.md@@ -45,30 +45,26 @@ messages = [{"role": "assistant","content": None,- "tool_calls": [{- "id": "call_123",- "type": "function",- "function": {"name": "search", "arguments": '{"q": "python"}'},- }],+ "tool_calls": [+ {+ "id": "call_123",+ "type": "function",+ "function": {"name": "search", "arguments": '{"q": "python"}'},+ }+ ],},{"role": "tool","tool_call_id": "call_123",- "content": json.dumps({- "results": [- {"title": f"Tutorial {i}", "score": 100-i}- for i in range(500)- ]- }),+ "content": json.dumps(+ {"results": [{"title": f"Tutorial {i}", "score": 100 - i} for i in range(500)]}+ ),},{"role": "user", "content": "What are the top 3?"},]# Headroom compresses 500 results to ~15, keeping highest-scoring items-response = client.chat.completions.create(- model="gpt-4o-mini",- messages=messages-)+response = client.chat.completions.create(model="gpt-4o-mini", messages=messages)# Check savingsstats = client.get_stats()@@ -187,13 +183,10 @@ print(f"Transforms: {plan.transforms}")response = client.chat.completions.create(model="gpt-4o",messages=[...],-# Override mode for this requestheadroom_mode="audit",-# Reserve more tokens for outputheadroom_output_buffer_tokens=8000,-# Keep last N turnsheadroom_keep_turns=5,)@@ -203,6 +196,7 @@ response = client.chat.completions.create(```pythonimport logging+logging.basicConfig(level=logging.INFO)# Now you'll see:diff --git a/wiki/shared-context.md b/wiki/shared-context.mdindex 72a9f133f..949180e9c 100644--- a/wiki/shared-context.md+++ b/wiki/shared-context.md@@ -28,10 +28,10 @@ Store content under a key. Compresses automatically using Headroom's full pipeli```pythonentry = ctx.put("findings", big_json_output, agent="researcher")-entry.original_tokens # 20,000-entry.compressed_tokens # 4,000-entry.savings_percent # 80.0-entry.transforms # ["router:json:0.20"]+entry.original_tokens # 20,000+entry.compressed_tokens # 4,000+entry.savings_percent # 80.0+entry.transforms # ["router:json:0.20"]```### `get(key, *, full=False)`@@ -39,9 +39,9 @@ entry.transforms # ["router:json:0.20"]Retrieve content. Returns compressed version by default, original with `full=True`.```python-compressed = ctx.get("findings") # 4K tokens+compressed = ctx.get("findings") # 4K tokensoriginal = ctx.get("findings", full=True) # 20K tokens-missing = ctx.get("nonexistent") # None+missing = ctx.get("nonexistent") # None```### `get_entry(key)`@@ -50,13 +50,13 @@ Get the full `ContextEntry` with metadata.```pythonentry = ctx.get_entry("findings")-entry.key # "findings"-entry.agent # "researcher"-entry.original_tokens # 20000+entry.key # "findings"+entry.agent # "researcher"+entry.original_tokens # 20000entry.compressed_tokens # 4000-entry.savings_percent # 80.0-entry.timestamp # 1710000000.0-entry.transforms # ["router:json:0.20"]+entry.savings_percent # 80.0+entry.timestamp # 1710000000.0+entry.transforms # ["router:json:0.20"]```### `keys()`@@ -69,11 +69,11 @@ Aggregated stats across all entries.```pythonstats = ctx.stats()-stats.entries # 3-stats.total_original_tokens # 60000+stats.entries # 3+stats.total_original_tokens # 60000stats.total_compressed_tokens # 12000-stats.total_tokens_saved # 48000-stats.savings_percent # 80.0+stats.total_tokens_saved # 48000+stats.savings_percent # 80.0```### `clear()`@@ -85,8 +85,8 @@ Remove all entries.```pythonctx = SharedContext(model="claude-sonnet-4-5-20250929", # For token counting- ttl=3600, # 1 hour (default)- max_entries=100, # Evicts oldest when full+ ttl=3600, # 1 hour (default)+ max_entries=100, # Evicts oldest when full)```@@ -113,11 +113,13 @@ from headroom import SharedContextctx = SharedContext()+def researcher_node(state):result = do_research()ctx.put("research", result)return {"research_summary": ctx.get("research")}+def coder_node(state):# Compressed summary in state, full details on demandfull = ctx.get("research", full=True)@@ -131,6 +133,7 @@ from headroom import SharedContextctx = SharedContext()+def compress_handoff(messages):for msg in messages:if len(msg.content) > 1000:@@ -138,6 +141,7 @@ def compress_handoff(messages):msg.content = ctx.get(msg.id)return messages+handoff(agent=coder, input_filter=compress_handoff)```diff --git a/wiki/strands.md b/wiki/strands.mdindex 585340c35..172e21b2b 100644--- a/wiki/strands.md+++ b/wiki/strands.md@@ -114,11 +114,13 @@ HeadroomStrandsModel supports Strands' structured output feature:```pythonfrom pydantic import BaseModel+class Analysis(BaseModel):severity: strroot_cause: strrecommendation: str+result = optimized.structured_output(Analysis, messages)```diff --git a/wiki/text-compression.md b/wiki/text-compression.mdindex 8fd4d743a..8656ea7e9 100644--- a/wiki/text-compression.md+++ b/wiki/text-compression.md@@ -136,10 +136,14 @@ elif detection.content_type == ContentType.PLAIN_TEXT:```pythonfrom headroom.transforms import (- detect_content_type, ContentType,- SearchCompressor, LogCompressor, TextCompressor+ detect_content_type,+ ContentType,+ SearchCompressor,+ LogCompressor,+ TextCompressor,)+def compress_tool_output(content: str, context: str = "") -> str:"""Application-level compression with explicit control."""detection = detect_content_type(content)@@ -166,7 +170,7 @@ Each compressor accepts configuration options:from headroom.transforms import SearchCompressor, SearchCompressorConfigconfig = SearchCompressorConfig(- max_results=50, # Keep up to 50 matches+ max_results=50, # Keep up to 50 matchespreserve_file_diversity=True, # Ensure different files representedrelevance_threshold=0.3, # Minimum relevance score to keep)diff --git a/wiki/transforms.md b/wiki/transforms.mdindex 30ea74f7f..b599b5164 100644--- a/wiki/transforms.md+++ b/wiki/transforms.md@@ -22,13 +22,13 @@ SmartCrusher analyzes JSON arrays and selectively keeps important items:from headroom import SmartCrusherConfigconfig = SmartCrusherConfig(- min_tokens_to_crush=200, # Only compress if > 200 tokens- max_items_after_crush=50, # Keep at most 50 items- keep_first=3, # Always keep first 3 items- keep_last=2, # Always keep last 2 items- relevance_threshold=0.3, # Keep items with relevance > 0.3- anomaly_std_threshold=2.0, # Keep items > 2 std dev from mean- preserve_errors=True, # Always keep error items+ min_tokens_to_crush=200, # Only compress if > 200 tokens+ max_items_after_crush=50, # Keep at most 50 items+ keep_first=3, # Always keep first 3 items+ keep_last=2, # Always keep last 2 items+ relevance_threshold=0.3, # Keep items with relevance > 0.3+ anomaly_std_threshold=2.0, # Keep items > 2 std dev from mean+ preserve_errors=True, # Always keep error items)```@@ -94,9 +94,9 @@ result = aligner.align(messages)from headroom import CacheAlignerConfigconfig = CacheAlignerConfig(- extract_dates=True, # Move dates to dynamic section- normalize_whitespace=True, # Consistent spacing- stable_prefix_min_tokens=100, # Min prefix size for alignment+ extract_dates=True, # Move dates to dynamic section+ normalize_whitespace=True, # Consistent spacing+ stable_prefix_min_tokens=100, # Min prefix size for alignment)```@@ -170,16 +170,16 @@ pip install "headroom-ai[code]" # Adds tree-sitter-language-packfrom headroom.transforms import CodeAwareCompressor, CodeCompressorConfig, DocstringModeconfig = CodeCompressorConfig(- preserve_imports=True, # Always keep imports- preserve_signatures=True, # Always keep function signatures- preserve_type_annotations=True, # Keep type hints- preserve_error_handlers=True, # Keep try/except blocks- preserve_decorators=True, # Keep decorators+ preserve_imports=True, # Always keep imports+ preserve_signatures=True, # Always keep function signatures+ preserve_type_annotations=True, # Keep type hints+ preserve_error_handlers=True, # Keep try/except blocks+ preserve_decorators=True, # Keep decoratorsdocstring_mode=DocstringMode.FIRST_LINE, # FULL, FIRST_LINE, REMOVE- target_compression_rate=0.2, # Keep 20% of tokens- max_body_lines=5, # Lines to keep per function body- min_tokens_for_compression=100, # Skip small content- language_hint=None, # Auto-detect if None+ target_compression_rate=0.2, # Keep 20% of tokens+ max_body_lines=5, # Lines to keep per function body+ min_tokens_for_compression=100, # Skip small content+ language_hint=None, # Auto-detect if None)compressor = CodeAwareCompressor(config)@@ -263,10 +263,10 @@ ContentRouter analyzes content and selects the best compression strategy:from headroom.transforms import ContentRouter, ContentRouterConfig, CompressionStrategyconfig = ContentRouterConfig(- min_section_tokens=100, # Minimum tokens to compress- enable_code_aware=True, # Use CodeAwareCompressor for code- enable_search_compression=True, # Use SearchCompressor for grep output- enable_log_compression=True, # Use LogCompressor for logs+ min_section_tokens=100, # Minimum tokens to compress+ enable_code_aware=True, # Use CodeAwareCompressor for code+ enable_search_compression=True, # Use SearchCompressor for grep output+ enable_log_compression=True, # Use LogCompressor for logsdefault_strategy=CompressionStrategy.TEXT, # Fallback strategy)@@ -331,10 +331,12 @@ Combine transforms for optimal results.```pythonfrom headroom import TransformPipeline, SmartCrusher, CacheAligner-pipeline = TransformPipeline([- SmartCrusher(), # First: compress tool outputs- CacheAligner(), # Then: stabilize prefix-])+pipeline = TransformPipeline(+ [+ SmartCrusher(), # First: compress tool outputs+ CacheAligner(), # Then: stabilize prefix+ ]+)result = pipeline.transform(messages)print(f"Saved {result.tokens_saved} tokens")diff --git a/wiki/troubleshooting.md b/wiki/troubleshooting.mdindex 35907f057..ca505fcb0 100644--- a/wiki/troubleshooting.md+++ b/wiki/troubleshooting.md@@ -224,6 +224,7 @@ client = HeadroomClient(# 2. For temp directory storageimport tempfileimport os+db_path = os.path.join(tempfile.gettempdir(), "headroom.db")client = HeadroomClient(original_client=OpenAI(),@@ -287,6 +288,7 @@ pip install --upgrade headroom-ai```python# Check available importsimport headroom+print(dir(headroom))# Common imports:@@ -397,6 +399,7 @@ print(f"Waste signals: {plan.waste_signals}")# See the actual optimized messagesimport json+print(json.dumps(plan.messages_optimized, indent=2))```