agent-skills

Analyze conversation to improve global ~/CLAUDE.md with cross-project preferences and behavior patterns. Use this when Claude repeated a mistake it makes everywhere, when you had to re-explain a preference that should apply to all projects, or when you notice a communication or workflow pattern worth capturing globally. Trigger with "reflect globally", "update global instructions", "capture cross-project preference", or "improve global CLAUDE.md". Also use when /reflection surfaces findings that aren't project-specific.

Analyze conversation to improve the project's CLAUDE.md. Use this at the end of any session, after correcting Claude, after explaining something twice, or when you notice Claude guessing about project conventions. Trigger with "reflect", "update CLAUDE.md", "capture what we learned", or "improve project instructions". Even if the session felt smooth, there may be implicit knowledge worth documenting.

Use when you need to recall past conversations, find how a problem was solved before, look up previous decisions, search for context from earlier sessions, or when the user asks about conversation history

Use when the user asks for a summary of work activity, daily/weekly standup report, what someone worked on, activity log, or needs to see all PRs, MRs, and Jira tickets touched during a specific date range

Use when the user asks to "address PR comments", "fix review comments", "handle PR feedback", or "respond to reviewers". By default operates on the current PR (from branch or conversation context). Use with "all" argument to process all open PRs with unanswered reviewer comments.

Use when the user asks to "update PR description", "update MR description", "analyze PR/MR changes and update comment", "summarize PR", "summarize MR", "generate PR/MR body from changes", or wants to automatically update a pull request or merge request description based on its code changes

Fix all code review issues found in the current conversation. Use when the user says "/fix", "fix all review issues", "fix the review findings", "apply the fixes", "fix all FAILs", or similar phrases after a code review has been run (e.g., after /review-prs, a code-reviewer agent output, or any structured findings list). Automatically parses FAIL and WARN findings from the conversation, applies fixes to local files, and commits the result. Also trigger if the user says "fix all found issues" or "fix everything from the review".

Migrate a project from Claude-specific instruction files (CLAUDE.md, .claude/skills/) to the universal AGENTS.md standard, creating symlinks so every configured AI agent shares one source of truth. Handles four starting states: CLAUDE.md only, AGENTS.md only, both (conflict), or neither (scaffold). Use this skill whenever the user wants to convert to agents.md, make project instructions universal or agent-agnostic, set up or adopt the agents.md standard, migrate from CLAUDE.md, share instructions across multiple AI tools (Augment, Claude, Gemini, Copilot, Cursor, Windsurf), or move .claude/skills to .agents/skills. Also triggers on phrases like "make instructions work for other agents", "new dev uses a different AI tool", or any mention of the agents.md spec. Do NOT use this skill when the user just wants to edit CLAUDE.md/AGENTS.md content, review existing instructions, or configure .claude/settings.json — those are content editing tasks, not migration.

Remove signs of AI-generated writing from text. Use when editing or reviewing text to make it sound more natural and human-written. Based on Wikipedia's comprehensive "Signs of AI writing" guide. Detects and fixes patterns including: inflated symbolism, promotional language, superficial -ing analyses, vague attributions, em dash overuse, rule of three, AI vocabulary words, passive voice, negative parallelisms, and filler phrases.

Use when the user wants to write, draft, or create a LinkedIn post — to announce a project, tool, skill, launch, blog, feature, milestone, or any subject they want to share professionally. Triggers on "write a LinkedIn post", "post this on LinkedIn", "draft a LinkedIn announcement", "make a LinkedIn post about X", or when someone has built/shipped something and wants to share it on LinkedIn. Produces a browser-openable HTML preview with copy buttons (plain text + Unicode-bold + first-comment) and a separate model-agnostic image-generation prompt as a .md file. Use even if they don't say the word "LinkedIn" but clearly want a professional social post for a dev/work audience.

Use when turning a GitHub user's PR activity into CV achievement bullets for a job entry. If given no data path, it fetches the PRs itself via the bundled fetch-prs.sh (needs gh + jq); if given a JSON file or directory it analyzes that instead. Processes both authored and reviewed PRs in one pass, prints the bullets, and saves them to a timestamped results file beside the data. Does not edit the CV itself.

Run sync.sh --check and pre-commit on a named plugin to surface validation errors before committing.

Scaffold a new plugin bundle in this marketplace — creates plugin.json, SKILL.md with correct frontmatter, per-plugin README, then runs sync.sh.