gzb1128
UserSkill Forge: Claude Code plugin marketplace for agent harness docs, code quality workflows, and OpenCode customization.
Categories
Indexed Skills (12)
curate
Use when the user says "curate", "audit docs", or wants to review the docs/ knowledge base for stale links, encyclopedia bloat, naming drift, and missing indexes — the docs/ counterpart to /agent-docs:remember
learn
Use when the user says "learn", "save this insight", or wants to persist valuable knowledge from the current session to AGENTS.md or docs/
remember
Use when the user says "remember", "audit knowledge", or wants to review and reorganize AGENTS.md memory for staleness, duplication, and misplacement
skill-creator
Create, migrate, adapt, test, and improve skills in Claude Code plugin repositories, especially skill-forge. Use when users ask to create or update a skill, migrate an upstream skill, compare official skills, design RED/GREEN verification, run skill evals, benchmark skill behavior, optimize triggering descriptions, or package a skill/plugin.
bootstrap-agent-docs
Use when a repository has no project AGENTS.md and needs a minimal agent entry point with verified commands and architecture routing, or when the user says "bootstrap agent docs", "init agent docs", or "scaffold AGENTS.md"
clean-commit
Use when the user asks to commit local code changes, or after a feature or bugfix is ready to land.
diff-cleanup
Use when the user asks to remove AI slop, clean up AI-generated code, strip bloated comments, or simplify code on a feature branch. Triggers on "remove slop", "clean up AI code", "remove unnecessary comments", "simplify this diff", "this code feels bloated".
loopfix
Use when an approved goal, design, or implementation plan needs autonomous hardening after a first pass, especially when the user says loopfix, loop-fix, review-fix loop, keep fixing, or do not wait unless impact is broad.
find-contributable-issues
Use when a contributor wants to find GitHub issues worth picking up — ranking open issues by how contribute-able they are with difficulty, claimed status, linked PRs, staleness, maintainer engagement, and area. Triggers on "good first issue", "what can I work on", "what's up for grabs", "any low-hanging fruit", "easy issues to pick up", "find a bug to fix", "issues I could take", "help wanted issues", or when a user is about to start contributing to a repo and wants to know what's worth claiming.
hydrate-opencode-models
Use when adding models to a custom opencode provider and needing to fill in limit, modalities, reasoning, tool_call, cost, interleaved, temperature, and attachment from the Models.dev catalog. Triggers on "fill model params", "hydrate model config", "add model metadata", "configure model specs", or when editing provider.models in opencode.json/opencode.jsonc.
integrate-projects
Use when a project needs persistent access to one or more external codebases in opencode. Triggers on "integrate projects", "add external projects", "link codebases", or "configure project references".
codex-subagent-strategy
Use in Codex when the user explicitly asks to spawn subagents, delegate work, or run parallel agents and the parent is about to dispatch them; read this strategy before the first spawn and route implementation workers by task shape. Also use it when the user requests parent-model inheritance for every child so the opt-out is enforced.
Bio shown is the top-scored skill's repo description as a fallback — real GitHub bios land in a future update.