cc-toolkit

Produce a source-grounded, cited HTML research report on a question. Gathers a brief (depth, recency, source mix, audience) then launches a bundled Dynamic Workflow that fans out firecrawl retrieval into findings, gates rounds on coverage + contradictions, synthesizes a cited answer, and renders an evolving HTML report. Use for substantive research requests: 'research X', 'zbadaj/zresearchuj X', 'do a deep dive on', 'find out everything about', 'write me a report on', 'compare A vs B with sources', 'what does the evidence say about'. Invoke explicitly as /researcher:research "<question>". Requires Dynamic Workflows enabled + the firecrawl MCP.

Code review focused exclusively on comment quality — not logic, naming, or structure. Use when the user says "/comment-review", "review komentarzy", "sprawdź komentarze", "przejrzyj komentarze", "comment review", or asks to check whether comments in a change are worth keeping. Reviews the current branch diff by default (auto-detects the base branch, or pass --base), or explicit file/dir paths when passed as arguments. Judges every comment against a focused rule set — no code-narration, decisions-only, not too long, no cross-file/doc/spec-id refs (file paths and bare requirement tokens like F1, Q1, R2, §4.1), no banner sections, no change-state/ticket history, no process-narration disguised as a decision, no commented-out code, and nothing that contradicts the code, and no rationale parked on the wrong declaration when it belongs where the behavior lives — and returns a per-comment verdict (KEEP / REMOVE / REWRITE / MOVE) with a concrete suggested fix.

Code review focused on code quality and craft — readability, structure, and style — not correctness, security, or naming. Use when the user says "/quality-review", "review jakości", "sprawdź jakość kodu", "przejrzyj pod kątem jakości", "quality review", "czy ten kod się dobrze czyta", or asks whether a change reads cleanly and is well structured. Especially apt for code freshly generated by a strong model, which is correct but glued together: blocks with no blank lines, helpers ordered randomly, OOP and functional styles mixed, pointless barrel exports, unnecessary casts, and two near-identical functions that should be one. Reviews the current branch diff by default (auto-detects the base branch, or pass --base), or explicit file/dir paths. Returns per-finding severity + a concrete suggested fix, and offers to apply the safe ones.

Orchestrate local-file transcription pipeline — transcribe audio/video files from disk and process them (summarize, extract domain news, or apply custom prompts). Use whenever the user provides paths, globs, or folders pointing at local audio/video files combined with analysis intent. Triggers on: paths ending with audio/video extensions (mp3, m4a, wav, mp4, mov, mkv, etc.) + 'transkrybuj', 'podsumuj nagranie', 'transcribe these files', 'extract action items', 'co jest w tym nagraniu', 'wyciągnij notatki z meetingów', OR folder/glob paths combined with the same intent keywords. Always use this — not local-transcribe or transcript-process separately — when the user wants end-to-end local file analysis. For YouTube URLs instead, use the `yt` skill.

Transcribe local audio/video files to Markdown via ffmpeg + ElevenLabs Scribe. Use whenever the user provides paths to local audio/video files (mp3, mp4, m4a, wav, mov, mkv, etc.) and wants raw transcription without summarization. Triggers on: paths with audio/video extensions + 'transcribe', 'transkrybuj', 'pobierz tekst z nagrania', 'get transcript of this file'. Sub-skill of `local` orchestrator — usually invoked via that. Use directly when the user wants only transcription, no processing.

Process audio/video transcriptions — summarize, extract news/novelties from a specific domain, or apply a custom prompt. Source-agnostic: works with transcripts from yt-transcribe, local-transcribe, or any other producer. Use whenever user has a transcript and wants to do something with it: 'summarize this', 'what's new in AI from this episode', 'extract key points', 'find news about X', 'process this transcript', 'wyciągnij action items', 'podsumuj nagranie'.

Orchestrate YouTube video analysis pipeline — transcribe and then process videos (summarize, extract domain news, or apply custom prompts). Use whenever user provides YouTube URLs and wants to extract information from them: summaries, news, novelties, key points, analysis, or custom processing. Triggers on: YouTube links + any analysis intent, 'co nowego w tych filmach', 'podsumuj filmy', 'summarize these videos', 'extract info from YT', 'what's new in AI from these episodes', 'przeanalizuj te filmy', or any request combining YouTube URLs with content extraction. Always use this — not yt-transcribe or transcript-process separately — when the user wants end-to-end YouTube video analysis. For local audio/video files instead of YouTube, use the `local` skill.

Download YouTube videos as audio and transcribe them to text using yt-dlp and ElevenLabs Scribe. Use whenever the user shares YouTube URLs/links and wants transcription, audio extraction for transcription, or text content from videos. Triggers on: YouTube links with transcription intent, 'transcribe this video', 'get text from YT', 'download and transcribe', 'what does this video say', or any request combining YouTube URLs with text extraction.

Use this skill when the user asks to "review changes against hookify rules", "check hookify compliance", "audit my diff for hookify violations", "run hookify code review", "review the PR for hookify rules", or "scan my changes for the hookify ruleset". Performs a hookify-rules-aware code review of committed and/or uncommitted changes by partitioning files into rule-scoped groups and dispatching parallel subagents, then aggregates a per-rule violation report and offers to save it or enter Plan Mode for fixes.

Orchestrate business name discovery: gather context, generate candidates, challenge them with a scoring sub-agent, then check domain availability. Use when the user asks to name a business, startup, app, or product — or wants brand name ideas. Trigger phrases: "help me name my business", "business name ideas for", "what should I call my company", "find a business name", "brand name generator", "suggest names for my startup", "name my app", "find an available domain name for my business". Also triggered by /namesmith. Examples: <example> user: "I'm building a project management SaaS, help me name it" assistant: "I'll use the namesmith skill to generate and evaluate name candidates, then check domain availability." <commentary>User describes a product and wants name ideas — namesmith triggers.</commentary> </example> <example> user: "I need a name for my pet care startup" assistant: "Let me run namesmith to brainstorm, score, and check domains for your pet care business." <commentary>Direct naming request for a

This skill should be used when the user asks to "create a hookify rule", "write a hook rule", "configure hookify", "add a hookify rule", or needs guidance on hookify rule syntax and patterns.

Analyze conversation history to extract development patterns, corrections, and deviations, then interactively update project documentation (.claude/rules/, CLAUDE.md). Use when finishing a development session, after resolving issues, or when user says "learn", "update rules", "session review", "retrospective", "upgrade knowledge", "capture learnings", or "what did we learn".

Run all configured linters across the repository and report issues with file paths and line numbers

Scaffold a new Claude Code plugin with all required files and register it in the marketplace

Validate plugin structure, marketplace.json consistency, and changelog format across all plugins