Relocate the script into the skill (skills/gitea/scripts/gitea_mcp.py) and add a
PEP 723 dependency block, so `uv run scripts/gitea_mcp.py` provisions deps with no
install step and the skill folder is portable (drop into ~/.cline/skills/). Drop
the non-standard `metadata` frontmatter — Cline only recognizes name + description.
- .mcp.json points at the relocated script and runs with --no-project (PEP 723)
- SKILL.md uses relative scripts/ paths; metadata block removed
- find_config walks up the tree to locate config.json (skill root or plugin root)
- tests import gitea_mcp via pytest pythonpath; README/CLAUDE paths updated
Tests: 119 passed, 92% coverage. Standalone list-tools verified via uv --no-project.
Expose the same 66 Gitea tools two ways from server/gitea_mcp.py:
- MCP server (no subcommand -> mcp.run()), as before
- CLI: `<tool> --flags` prints one JSON object {success, data/error}
A new @tool decorator registers each function in both FastMCP and a TOOLS
registry; a CLI dispatcher builds argparse from each function's signature and
wraps results in the repo's JSON+success contract. Adds `list-tools` for
discovery. No tool logic duplicated.
- Rewrite skills/gitea/SKILL.md for CLI-first usage (MCP kept as an option)
- Sync version to 0.3.0 across plugin.json, pyproject.toml, SKILL.md
- Add tests/unit/test_cli.py (dispatch routing, flag typing, errors, main)
- Document dual-mode in README.md and CLAUDE.md
- Ignore coverage artifacts
Tests: 119 passed, 93% coverage.
- Full Gitea REST API coverage for repos, issues, PRs, branches,
files, releases, webhooks, orgs, notifications
- Multi-instance support via config.json
- pyproject.toml with pinned dependencies (mcp[cli], httpx)
- config.example.json as credentials template (config.json is gitignored)
a-limasov-ii