{
  "markdown": "# skill-kit\n\nAgent skills for the points where an agent's work reaches a person.\n\n| When | Skill | What it does |\n|---|---|---|\n| A decision is needed | `quiz` | Asks it as multiple choice |\n| Options to pick from | `variant-lab` | Builds and tests 4 to 6 working versions |\n| Writing is handed over | `writing-check` | Flags AI writing patterns |\n| A graded or shared document | `deliverable` | Drafts it from the rubric |\n| The agent says done | `proof-before-done` | Shows the tool result as proof |\n| A push to a public repo | `push-check` | Blocks secrets, malicious code and personal information |\n| A session or tool ends | `handoff` | Writes and picks up handoff notes |\n\nThese are the points where agent work costs a person time or embarrassment: a question buried\nin a long reply, a claim with no source, a key in a public commit. Process kits such as\n[superpowers](https://github.com/obra/superpowers) cover how an agent builds. This kit covers\nwhat happens when that work meets a person, and works alongside them.\n\nEach skill is a plain `SKILL.md` folder, so any agent tool that reads that format can use it.\n\n## Install\n\n### Claude Code\n\n```\n/plugin marketplace add S3yon/skill-kit\n/plugin install skill-kit@skill-kit\n```\n\nThe skills then show as `skill-kit:<name>`, for example `skill-kit:quiz`.\n\nChecked against https://code.claude.com/docs/en/plugin-marketplaces, 2026-10-02.\n\n### GitHub Copilot CLI\n\nReads the same plugin manifests as Claude Code.\n\n```\ncopilot plugin marketplace add S3yon/skill-kit\ncopilot plugin install skill-kit@skill-kit\n```\n\nOr in one step: `copilot plugin install S3yon/skill-kit`.\n\nChecked against https://docs.github.com/en/copilot/reference/copilot-cli-reference/cli-plugin-reference, 2026-10-02.\n\n### Qwen Code\n\nReads the same marketplace file as Claude Code.\n\n```\nqwen extensions install S3yon/skill-kit:skill-kit\n```\n\nChecked against https://github.com/QwenLM/qwen-code/blob/main/docs/users/extension/introduction.md, 2026-10-02.\n\n### Gemini CLI\n\nReads `gemini-extension.json` and loads `skills/`.\n\n```\ngemini extensions install https://github.com/S3yon/skill-kit\n```\n\nChecked against https://github.com/google-gemini/gemini-cli/blob/main/docs/extensions/reference.md, 2026-10-02.\n\n### Codex\n\nReads the root `plugin.json` and finds `skills/` on its own.\n\n```\ncodex plugin marketplace add S3yon/skill-kit\ncodex plugin add skill-kit@skill-kit\n```\n\nChecked against https://developers.openai.com/plugins/build/plugins, 2026-10-02.\n\n### Cursor\n\nReads the root `plugin.json` and `.cursor-plugin/marketplace.json`. Open Customize, then\nFrom GitHub Repository, and enter `S3yon/skill-kit`.\n\nChecked against https://cursor.com/docs/plugins, 2026-10-02.\n\n### Copy a folder\n\nCopy one folder from `skills/` into the tool's skills folder:\n\n| Tool | For your user | For one project |\n|---|---|---|\n| Claude Code | `~/.claude/skills/` | `.claude/skills/` |\n| Codex, Cursor, Gemini CLI, Copilot CLI | `~/.agents/skills/` | `.agents/skills/` |\n| Qwen Code | `~/.qwen/skills/` | `.qwen/skills/` |\n\nNo manual install test was run. Of these tools, only the Cursor CLI was installed on the\nmachine where the steps were checked, and it was not signed in, so it could not load a plugin.\n\n## Always-on snippet\n\nSkills load when their description matches the task. To have three of the rules apply on every\ntask, paste these lines into a project's agent instructions file:\n\n```\n- Report an action as done only with the tool result that shows it, and an outside fact only with its source.\n- Put any draft that needs approval in full in the reply text.\n- Run push-check before any push to a public remote.\n```\n\n## The scripts on their own\n\nBoth need only Python 3.10 or later. `push_check.py` also needs git.\n\n`skills/writing-check/scripts/slop_check.py` checks a draft:\n\n```\npython3 skills/writing-check/scripts/slop_check.py draft.md\n```\n\nIt prints one line per flag with the line number and a replacement. Exit 0 means clean, 1 means\nsomething was flagged, 2 means a file could not be read. The patterns and why they are on the list:\n[tells.md](skills/writing-check/tells.md).\n\n`skills/push-check/scripts/push_check.py` checks commits before a push:\n\n```\npython3 skills/push-check/scripts/push_check.py --remote origin\npython3 skills/push-check/scripts/push_check.py --install-hook\n```\n\nWith no range it scans the unpushed commits. `--install-hook` installs it as the repo's\npre-push hook. Exit 0 means nothing blocks, 1 means something does, 2 means a usage error.\n\n## Tests, CI and evals\n\n- `python3 -m unittest discover tests` runs the script tests.\n- `python3 scripts/validate.py` checks every skill folder, eval file and plugin manifest.\n- CI runs both on every push and pull request, on Python 3.10 and the latest 3.x.\n- Each skill's `evals/evals.json` holds prompts with checkable assertions. How to run and grade\n  them: [docs/evals.md](docs/evals.md).\n\n## License\n\nMIT\n",
  "bytes": 4901,
  "sha": "92bf16afee90cfdb058f0c21681e9758286b6c9f56a8ca0b480576a8f0e1fe53",
  "repo_slug": "s3yon/skill-kit",
  "fonte": "repo",
  "truncated": false,
  "api": "https://api.agentalog.com/api/listings/plg_s3yon_skill_kit_c08184eb/readme"
}