メニュー
Step-by-step release workflow for publishing pycode-kg to PyPI using Poetry. Use this skill when the user wants to cut a release, bump the version, update the CHANGELOG, tag a commit, build and publish to PyPI, or run any part of the release pipeline for the pycode-kg project.
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
SOC 職業分類に基づく
Expert knowledge for using AgentKG — conversational memory as a live, queryable knowledge graph. Use this skill when: working in a repo that has AgentKG installed (check for a .agentkg/ directory or agent-kg in pyproject.toml), querying past conversation context, understanding what a user has told you before, surfacing user preferences or commitments, running or debugging agent-kg CLI commands (agentkg-ingest, agentkg-query, agentkg-assemble, agentkg-prune, agentkg-stats, agentkg-analyze, agentkg-sessions, agentkg-snapshot, agentkg-onboard, agentkg-profile, agentkg-viz, agentkg-wipe, agentkg-mcp), setting up hooks to auto-ingest conversation turns, managing the UserProfile, or troubleshooting why profile or memory data is missing.
Expert knowledge for installing, configuring, and using DocKG — a hybrid semantic + structural knowledge graph for document corpora (.md and .txt files). Use this skill when the user asks about: setting up DocKG in a project, adding doc-kg as a Poetry dependency, building the SQLite or LanceDB knowledge graph from documents, running the multipass analysis pipeline (dockg pipeline run/embed/manifold), configuring .mcp.json for Claude Code or Kilo Code, configuring .vscode/mcp.json for GitHub Copilot, configuring claude_desktop_config.json for Claude Desktop, using the dockg CLI (dockg build, dockg build-graph, dockg build-index, dockg query, dockg pack, dockg analyze, dockg semantic-analyze, dockg pipeline, dockg viz, dockg mcp, dockg snapshot), using the graph_stats / query_docs / pack_docs / get_node MCP tools, or troubleshooting DocKG errors.
Guide for creating effective skills. This skill should be used when users want to create a new skill (or update an existing skill) that extends Claude's capabilities with specialized knowledge, workflows, or tool integrations.
| name | publish |
| description | Step-by-step release workflow for publishing pycode-kg to PyPI using Poetry. Use this skill when the user wants to cut a release, bump the version, update the CHANGELOG, tag a commit, build and publish to PyPI, or run any part of the release pipeline for the pycode-kg project. |
git status shows no uncommitted changes)poetry run pytest)pre-commit run --all-files)poetry config pypi-token.pypi <token>)Follow Semantic Versioning:
| Change | Bump |
|---|---|
| Bug fixes only | patch (0.5.2 → 0.5.3) |
| New features, backward-compatible | minor (0.5.2 → 0.6.0) |
| Breaking changes | major (0.5.2 → 1.0.0) |
Move items from ## [Unreleased] to a new versioned section:
## [X.Y.Z] - YYYY-MM-DD
Leave ## [Unreleased] empty at the top for future entries.
poetry version patch # or minor / major
Verify: grep '^version' pyproject.toml
git add pyproject.toml CHANGELOG.md
PYCODEKG_SKIP_SNAPSHOT=1 git commit -m "chore: release vX.Y.Z"
Use
PYCODEKG_SKIP_SNAPSHOT=1to prevent the post-commit hook from creating a snapshot on a release commit.
git tag -a vX.Y.Z -m "Release vX.Y.Z"
poetry build
Artifacts appear in dist/. Verify they look correct:
ls -lh dist/
poetry publish
To test against TestPyPI first:
poetry publish -r testpypi
git push origin main --tags
.claude/skills/pycodekg/SKILL.md if the install command or version references changedpycodekg snapshot save X.Y.Z --repo . --commit $(git rev-parse HEAD) --branch mainIf something went wrong after poetry publish:
poetry run twine yank pycode-kg X.Y.Zgit revert HEAD then git pushgit tag -d vX.Y.Z && git push origin :refs/tags/vX.Y.Z