Menu
SkillsMP has collected 12 skills from fedemelo/claude-config. Open a skill to review its source and details.
Showing 12 of 12 collected skills.
Summarizes the user's own pull request activity in the current repo from a cutoff date to now: PRs still open, PRs merged, and PRs closed without merging, each with a one-line summary of what it does plus review and CI state. Use when asked what work was…
Triages PR review feedback (human comments and bot findings) on the current branch's PR or a given number or URL, fixes the real issues by following the commit skill, and returns replies in this session for the rest. Never pushes or posts to the PR. Use when…
Commits staged or pending changes following personal conventions: modular commits, terse single-line messages, no AI attribution. Use when asked to commit.
Pushes local commits to the tracked branch with the git-land tool, which wraps them in a disposable, auto-merged PR. Use when asked to land, ship, or merge local commits without a manual PR.
Reviews a GitHub PR as a staff engineer and returns numbered, categorized comments plus an APPROVE, COMMENT, or REQUEST CHANGES verdict in this session, without posting to the PR or changing code. Takes an optional PR number or URL (defaulting to the current…
Puts up a clean PR for the current working changes: cuts a fresh branch off the latest default branch, commits only valid code changes, pushes, and opens a PR assigned to the user with a terse description. Use when asked to open, put up, raise, or submit a PR.
Gives an informed opinion on a comment someone left on a PR: refetches so the view is current, works out which comment is meant without being told, checks the claim against the code, and says plainly whether it holds. Never changes code, posts, replies, or…
Files a GitHub issue with the git-todo tool, auto-assigned to the user, for lightweight personal backlog tracking. Use when asked to create a todo or issue, or track a bug or task on GitHub.
Audits the discussion on a PR after the replies have been posted, checking every claim in them against the code and finding points that were raised but never answered. Flags what is wrong, partial, missed, imprecise, or left dangling, and changes nothing.…
The standard for prose meant for another person to read: B2-level English, no jargon or corporate fluff, no reference chains, nothing that reads as AI-written. Other skills reference it when writing review comments, replies, and descriptions. Use when writing…
The standard for PR descriptions: succinct, unformatted, at most two short paragraphs covering what was fixed and any non-obvious decisions. Other skills reference it when opening a PR or judging an existing description. Use when writing or reviewing a PR…
Defines which code comments are worth keeping: only those carrying knowledge that cannot be inferred from the code, not restatements of it. Strips unnecessary comments from a diff or file when invoked directly; other skills reference it as the rule for…