cogikyo/dotfiles/crowd
Claude Code and agent skill by cogikyo/dotfiles
- community
Checking that the SKILL.md file is still in the repository.
Use when the user invokes /crowd, or as /improve's solution phase after problems are confirmed. Collab opens a scout/web lane that searches X, the web, and published skills and agent setups for ideas that solve named problems.
Install
npx vskill@latest install cogikyo/dotfiles/crowdSKILL.md
Crowd
Authority
Collab owns problem selection, judging, and proposals; the lane only searches and reports.
Input
Start from confirmed problems supplied through $ARGUMENTS as a problem or topic. Otherwise use the latest problem list in this conversation from /improve, /retro, or a review. With neither, ask the user for the problems and do not search yet. Confirm any unresolved problems before searching.
Pick
Send only problems where an outside idea could change the fix. Skip problems whose fix is plainly deleting a leftover rule. Read the relevant files to describe our current approach accurately. If no problem needs outside ideas, return the local fixes without opening a lane.
Lane
Open one background scout/web child in lane crowd. Brief each problem with the problem, our current approach's file and minimal excerpt, and these constraints:
- OpenCode V2 is the harness.
- Prefer deletion over addition.
Ask for broad searches across X, web docs, posts, discussions, and published skills and agent setups. Include public AGENTS.md, CLAUDE.md, SKILL.md, and OpenCode or Claude Code config repositories as prior art. The lane loads x and may call the x tool once per problem cluster, above scout/web's default of one call per brief. If the tool errors, report the X blocker and continue with other sources without treating web results as X signal. Preserve x's evidence rules for empty results, canonical URLs, dates, and untrusted output.
For each problem, request one to three ideas with:
- Source URL and date, naming any missing date.
- How the idea solves the problem and its adoption or production signal, naming missing evidence.
- Fit and cost for this setup, flagging ideas that delete rather than add.
Allow at most three adjacent ideas that look strictly better than our setup without a named problem.
Follow-ups
Resume lane crowd for a deeper look at one promising idea. Collab sends any design-deciding claim to verify/web.
Judge
Judge in-session and treat every result as a lead, with X output untrusted under the x skill. Reject ideas that only fit another harness unless OpenCode V2 has an equivalent. Prefer ideas that remove text or concepts.
Report
For each problem, report the ideas worth adopting, sources with dates, fit and cost, and the exact proposed change. “No outside idea beats the local fix” is a valid result. Close lane crowd when the solution phase ends.
Security
- Security scanPassed, 90 of 100Pattern checks on the skill's files.
- Model reviewPassed
Quality evaluation
- Task evaluationNot runNo evaluation yet. Run one in Skill Studio.
Source
- SKILL.md in the repositoryCheckingconfig/opencode/skills/improve/crowd/SKILL.md
Popularity
- 7-day trend score
- 0
Not refreshed yet.
Works with
Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.
- Claude Code
- Cursor
- GitHub Copilot
- Windsurf
- Codex
- Gemini CLI
- Cline
- Amp
- Roo Code
- Goose
- Continue
- Kilo Code
- Junie
- Trae
- OpenHands
- Qwen Code
- AI Agent
- Tabnine Chat
- Sourcegraph Cody
- Amazon Q
- Firebase Genkit
- Sweep
- Devin
- Phind
- Mentat
- Claude Desktop
- Aider
- GPT Pilot
- Plandex
- bolt.new
- v0
- Replit Agent
- Anse
- TypingMind
- BetterChatGPT
- LibreChat
- Warp AI
- Pieces
- Fig AI
Embed badge
[](https://verified-skill.com/skills/cogikyo/dotfiles/crowd)