一键导入
add-skill
Import a skill from exactly one GitHub file URL, raw GitHub URL, or single-file gist URL into ~/.agents/skills/. Use when the user wants to add a skill from GitHub content.
菜单
Import a skill from exactly one GitHub file URL, raw GitHub URL, or single-file gist URL into ~/.agents/skills/. Use when the user wants to add a skill from GitHub content.
Address all review comments on a GitHub pull request. Fixes code, pushes back with evidence, resolves threads after completion, and updates AGENTS.md only when specifically requested. Use after receiving PR review feedback.
Tidy up CaPP git worktrees whose PRs are merged or closed without merging. Scans for directories matching the `<repo>_<TICKET-ID>` worktree convention (e.g. `twinkl-sanity-cms_CAPP-1234`), looks up each worktree's PR via `gh`, classifies by state, flags closed-unmerged PRs, and removes worktrees only after explicit user confirmation. Use when the user asks to clean up worktrees, prune CAPP worktrees, or tidy `~/Development`.
Create well-structured Jira tickets through adaptive interview. Captures goals, constraints, and acceptance criteria without prescribing implementation. Use when creating new work items for the CaPP team.
Implement a Jira ticket by iteratively completing subtasks using TDD. Creates a feature branch, one commit per subtask, and a draft PR. Also handles small well-defined bugs without subtasks. Use when ready to start coding.
Diagnose and fix GitHub Actions CI failures on a pull request. Fixes code issues, validates locally using identified check tiers, and pushes focused fixes. Use when CI is failing on a PR.
Centralised CAPP Jira metadata — cloud ID, field IDs, transition IDs, issue type IDs, link types. Invoke before any Jira MCP call to skip redundant discovery API calls and ensure consistent field references.
| name | add-skill |
| description | Import a skill from exactly one GitHub file URL, raw GitHub URL, or single-file gist URL into ~/.agents/skills/. Use when the user wants to add a skill from GitHub content. |
Import a Copilot skill from a GitHub-hosted text file into this repository.
https://github.com/<owner>/<repo>/blob/<ref>/<path>https://raw.githubusercontent.com/<owner>/<repo>/<ref>/<path>gist.github.com or raw gist URLs on gist.githubusercontent.comCreate the new skill immediately under:
~/.agents/skills/<normalized-name>/SKILL.md
Then report:
name:description:Do not dump the whole file unless the user asks.
Prefer the most direct structured fetch available in the current environment.
For repository files:
blob URLs, convert to the corresponding raw URL (raw.githubusercontent.com) before fetching, or use a GitHub file-content tool if one is available.For gists:
Only continue if the fetched content is text. If the content is binary, inaccessible, or missing, fail with a clear explanation.
Pick the destination skill name in this order:
name: from the fetched file, if presentNormalize the chosen value to lowercase kebab-case (replace spaces and underscores with -, drop non-alphanumeric characters).
Use that same normalized value for both:
name: field in SKILL.mdTarget directory: ~/.agents/skills/<normalized-name>/
If that directory already exists, stop and refuse. Do not overwrite, merge, or partially update it. Tell the user to choose a different name or remove the existing skill first.
SKILL.mdSource already has YAML front matter:
name: to the normalized kebab-case valuedescription: if presentdescription: only when it is missing (see step 6)Source has no YAML front matter:
name: and description: (see step 6)Body rules:
--- front matter delimiter.description:When a description must be synthesized, infer a short, concrete, action-oriented summary from the imported content that describes when the skill should be used. Match the style of existing skills: one sentence ending with a trigger phrase like "Use when…".
~/.agents/skills/<normalized-name>/SKILL.md.Created: ~/.agents/skills/<normalized-name>/SKILL.md
name: <normalized-name>
description: <final description>