Add Ruff, pre-commit hooks, and incremental lint checks in CI

[QuanCheng-QC]

Summary

This PR introduces a minimal Ruff-based quality gate for the Python codebase.

Changes

• replace black and flake8 in dev extras with ruff
• add baseline Ruff configuration in pyproject.toml
• add .pre-commit-config.yaml with ruff-check and ruff-format
• add a dedicated lint job to pytest.yml

CI behavior

• linting now runs only on changed Python files in the current PR/push
• if no Python files are changed, the lint job is skipped successfully
• this avoids blocking adoption on existing repo-wide lint/format debt

Scope

This PR intentionally does not:

• perform a repo-wide reformat
• enforce mypy in CI
• fix historical lint debt
• change business logic
  Fixes [DX] Add Python code formatting & linting infrastructure (ruff, pre-commit, CI) #318

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated (UTC)
openagents-studio	Ready	Preview, Comment	Mar 28, 2026 9:27am
openagents-workspace	Ready	Preview, Comment	Mar 28, 2026 9:27am

[JiwaniZakir]

The CI lint job installs ruff without a version pin (pip install ruff), while .pre-commit-config.yaml pins rev: v0.6.9. This means CI could run a newer Ruff version that rejects code the pre-commit hook accepts (or vice versa), creating inconsistent feedback between local dev and CI. Consider installing via pip install "ruff==0.6.9" or deriving the version from pyproject.toml (e.g., pip install -e ".[dev]" filtered to just ruff).

The workflow_dispatch event is listed in the on: trigger at the top of the file, but the "Collect changed Python files" step falls through to the else branch and silently exits with has_python=false, meaning manually-triggered linting is a no-op. That's likely unintentional — a workflow_dispatch run should probably lint all Python files or at least warn that it's unsupported.

The checkout uses fetch-depth: 0 (full history), but for the PR case immediately follows with git fetch --no-tags --prune --depth=1 origin "$base_ref", which only needs the tip of the base branch. A shallow clone (fetch-depth: 1) at checkout with an explicit deepen for the merge-base would be faster on repos with long history.