WooDragon's CC plugin & skill marketplace.
Plugins:
claude plugin add WooDragon/cc-pluginsSkills:
npx skills add WooDragon/cc-plugins -g| Plugin | Description |
|---|---|
| plan-review | Adversarial plan review via cross-model consultation (agy/Claude) |
| doc-gate | Document editing governance — skill gate + lexical recall advisory + link graph tools |
| ppt-press | Self-contained PPT publishing — scaffold + create + deploy + manage |
| code-search | Code search & symbol navigation — pick the right tool by search intent |
| deep-research | Deep research framework — 7-Stage pipeline + role-specialized subagents + multi-model harvest & citation-verification gate |
| pr-review | AI review for open GitHub PRs — grok CLI local synchronous review (default, multi-round follow-up) + Copilot bot asynchronous review (optional) |
| Skill | Plugin | Description |
|---|---|---|
| ppt-init | ppt-press | Scaffold a new Astro-based PPT framework project from scratch |
| ppt-create | ppt-press | Generate editorial magazine × e-ink web presentations (Astro) |
| ppt-deploy | ppt-press | Build, test, and deploy PPT decks to AWS Amplify |
| ppt-manage | ppt-press | List, search, and retrieve PPT deck URLs |
| doc-maintenance | doc-gate | Document maintenance workflow with pre/post-flight checks |
| code-search | code-search | Pick the right search tool (ctags/ast-grep/grep) by intent |
| deep-research | deep-research | Router skill for the 7-stage research pipeline + quality gates |
| pr-review | pr-review | Review an open PR by number — grok (default) or Copilot bot backend |
Each plugin's own README has full detail (env vars, architecture, test suites). This is the minimal path to trying each one.
plan-review — fully automatic, no invocation needed. Enter plan mode as usual; when Claude calls ExitPlanMode, the plugin intercepts the plan for adversarial cross-model review before it reaches you.
doc-gate — also automatic. Editing any .md file triggers the gate; if doc-maintenance hasn't been invoked this session, Claude is prompted to call it first. To use the workflow directly: ask Claude to "创建/更新文档" and it will invoke doc-maintenance and follow the pre/post-flight checklist.
# standalone link-graph queries, no skill invocation needed
python3 plugins/doc-gate/tools/docs-graph.py checkppt-press — scaffold → create → deploy, driven entirely through conversation:
mkdir my-ppt && cd my-ppt
# tell Claude Code: "初始化 PPT 项目" → ppt-init scaffolds the framework
# tell Claude Code: "帮我做一个关于 XX 的 PPT" → ppt-create generates the deck
# tell Claude Code: "部署上线" → ppt-deploy builds, tests, and ships itcode-search — nothing to invoke; it's a pure methodology skill that activates whenever a request looks like "find X in the codebase" (definition lookup, caller search, AST-shape match).
deep-research — scaffold a research project, then let the skill drive the pipeline:
bash "${CLAUDE_PLUGIN_ROOT}/scripts/create-research-project.sh" "研究主题" --type web-research
# then ask Claude Code a research question in that directory — the deep-research skill
# auto-triggers and orchestrates harvester/analyst/reviewer subagents through the 7 stagespr-review — check out the PR's branch, then ask Claude Code to "评审 PR 123" / "grok review PR 123". The grok backend prints the review to your terminal and keeps a session so follow-up rounds only send the incremental diff; the Copilot backend is opt-in and posts back onto the PR instead.