pyproject.toml

[project]
name = "coder-agent"
version = "0.1.0"
description = "A terminal coding agent: plans a change, edits files, runs tests, and iterates. Built with LangGraph, MCP, and local RAG."
readme = "README.md"
license = "MIT"
license-files = ["LICENSE"]
authors = [{ name = "Chaimaa Abi" }]
keywords = ["coding-agent", "langgraph", "mcp", "rag", "llm"]
requires-python = ">=3.11"
dependencies = [
    "langchain>=0.3",
    "langchain-core>=0.3",
    "langgraph>=0.4",
    "langchain-groq>=0.3",
    "langchain-google-genai>=2.1",
    "langsmith>=0.3",
    "pydantic>=2",
    "pydantic-settings>=2",
    "python-dotenv>=1",
    "typer>=0.15",
    "rich>=13",
    "mcp>=1.10",
    "langchain-mcp-adapters>=0.1",
    "pathspec>=1.1.1",
    "tree-sitter>=0.26.0",
    "tree-sitter-language-pack>=1.17.0",
    "chromadb>=1.0",
    "fastembed>=0.5",
    "langgraph-checkpoint-sqlite>=3.1.1",
    "httpx>=0.27",
    "langchain-openai>=1.6.2",
]

[project.optional-dependencies]
dev = [
    "pytest>=8",
    "pytest-asyncio>=0.24",
    "ruff>=0.6",
    "matplotlib>=3.8",
    "pytest-cov>=6",
    # scripts/demo.py draws the README GIF; matplotlib pulls Pillow in anyway, this makes it explicit.
    "pillow>=10",
    "langchain-ollama>=0.3",
]
# Only `evals/figures.py` draws; the agent itself never imports matplotlib.
figures = ["matplotlib>=3.8"]
# Only the docs site build (mkdocs.yml, .github/workflows/docs.yml) needs this.
docs = ["mkdocs-material>=9"]
# Third provider: a model served by a local Ollama daemon (`ollama:qwen2.5-coder:7b`). Not a core
# dependency because most users never run one; the code path is exercised in CI through `dev`.
ollama = ["langchain-ollama>=0.3"]

[project.scripts]
coder = "coder_agent.cli:app"

[build-system]
requires = ["hatchling"]
build-backend = "hatchling.build"

[tool.hatch.build.targets.wheel]
packages = ["src/coder_agent"]

[tool.ruff]
line-length = 100
target-version = "py311"
# Benchmark fixtures are the agent's target repos; one contains a planted bug on purpose.
extend-exclude = ["evals/suite", "evals/humaneval"]

[tool.pytest.ini_options]
testpaths = ["tests"]
asyncio_mode = "auto"

[tool.coverage.run]
source = ["coder_agent"]
# The MCP server is spawned as a subprocess by its tests, so its lines are not seen in-process;
# the number under-reports that module rather than paying for subprocess coverage hooks.

[tool.coverage.report]
precision = 0