From whipforaweeb-skills
Wire a project to use CodeGraph MCP (@colbymchenry/codegraph) alongside graphify. Verifies CLI/index/MCP/permissions, auto-runs `codegraph index` if index is missing, writes or removes the managed CLAUDE.md role-split block. Handles unlink when args contain unlink/ยกเลิก/cancel/remove/ลบ/ถอด.
How this skill is triggered — by the user, by Claude, or both
Slash command
/whipforaweeb-skills:codegraph-linkThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Verifies CodeGraph prerequisites, auto-builds the index if missing, and writes a guidance block to the project's `CLAUDE.md` so the agent prefers CodeGraph MCP over grep/Explore for code navigation. Does not install CodeGraph or edit `.claude.json`.
Verifies CodeGraph prerequisites, auto-builds the index if missing, and writes a guidance block to the project's CLAUDE.md so the agent prefers CodeGraph MCP over grep/Explore for code navigation. Does not install CodeGraph or edit .claude.json.
Root = cwd unless a path-like arg overrides it.
Strip the managed block (markers inclusive, regex (?s)<!-- codegraph-link:start -->.*?<!-- codegraph-link:end -->) from $root\CLAUDE.md. If the file or block is absent, say "nothing to unlink". Leave .codegraph/, MCP config, and permissions intact (point user to codegraph uninstall). Confirm in one line, then stop.
where.exe codegraph. Missing → stop, tell user to install CodeGraph (npm i -g @colbymchenry/codegraph) then codegraph init.codegraph status -j $root → parse JSON. initialized:true → OK. initialized:false (or non-zero exit, or no .codegraph\) → run codegraph init in $root (init + initial index runs by default as of 0.9.7+; the old -i flag is deprecated but still accepted); show output; non-zero exit → stop, report error. If .codegraph\ exists but status reports not-initialized/empty (corrupt or interrupted index), run codegraph index to rebuild.codegraph entry in ~/.claude.json.
codegraph install --target=claude -l global -y (non-interactive; also writes the auto-allow list). Do not hand-edit .claude.json.codegraph install --print-config claude and compare its command/args to the live entry. Differ (stale config after a CLI upgrade) → report the drift and suggest re-running the install line above to refresh."env": { "CODEGRAPH_MCP_TOOLS": "explore,search,node,callers,callees,impact,files,status" } to the entry to expose only the tools you use (comma short-names; drop ones you never call to shrink the per-session tool schema).cursor, codex, opencode, hermes, gemini, antigravity, kiro — one index, many agents via codegraph install -t <ids> -y.mcp__codegraph__* allow entries in ~/.claude/settings.json. Canonical set (8 tools): codegraph_explore, codegraph_search, codegraph_callers, codegraph_callees, codegraph_impact, codegraph_node, codegraph_files, codegraph_status.$root\graphify-out\graph.json: exists → blocks/graphify.md, else → blocks/code-only.md (and note user can rerun after /graphify to upgrade). Read that block file (it already contains the markers) and write it verbatim into $root\CLAUDE.md (UTF-8 no BOM, create if missing). Idempotent: if the marker regex above matches, replace that span; else append. Use a literal replacement so $/regex metachars survive.codegraph-link: <full split | code-only> block written to <root>\CLAUDE.md; index OK, graphify <present/absent>, MCP <present/missing/drift>, auto-allow <present/missing>.codegraph serve --mcp..codegraph/ index.npx claudepluginhub vveb1250/whipforaweeb-skills --plugin whipforaweeb-skillsFetches up-to-date documentation from Context7 for libraries and frameworks like React, Next.js, Prisma. Use for setup questions, API references, and code examples.
Applies a firm's KYC/AML rules grid to parsed onboarding records: assigns risk rating, checks required documents, outputs rule outcomes with citations, and routes for escalation.
Generates daily or weekly digests of activity from connected sources (chat, email, docs, tasks, CRM), highlighting action items, decisions, mentions, and project updates.