hooks plugins

by Ivan-WG

Candour as a mechanism, not a pep talk. One blocking Stop gate, five clauses a script can prove: a file:line citation resolving to no file or past its end; a position retracted after bare pushback with no tool call in between; a completion claim after edits with nothing executed since; a task-runner run stopping without its recorded gate pass; a dependency map changed with an untouched lockfile. Flattery, apologies and defensive phrasing are counted by /candor:check and blocked by nothing, on purpose. Plus the terse reply mode (/candor:level lite|full|ultra|off): chat-message brevity as a shape contract — prose-line budgets and a fixed report skeleton, never less work in the turn.

by Colton Bearden

Agentic file librarian for carrel desks: /index, /doctor, /catalog, /completion, a file-librarian agent, an agent-workflows skill, the carrel MCP server (14 tools + carrel:// resources), and a PostToolUse hook that re-indexes files Claude writes.

by Colton Bearden

Hooks that let Claude read what it otherwise cannot: a PreToolUse guard converts PDFs, Office/ebook files and images to text via carrel before Read sees them, and a SessionStart hook reports what carrel can do in this environment.

by Thanh Nguyen

Lint edited JavaScript and TypeScript files for maintainable comments and god-file limits with the project's local ESLint installation.

by Ivan-WG

Destructive-command prevention: a PreToolUse hook that classifies every Bash (and MCP shell/SQL) command before it runs — denying irreversible data loss (migrate:fresh, db:wipe, DROP DATABASE, rm -rf /, git clean -fdx, docker compose down -v, terraform destroy, cloud deletes) and prompting on scoped-but-destructive ones (git reset --hard, rm -rf ./dir, kubectl delete pod). Reads through quotes, wrappers and heredocs; skips read-only commands; fail-open. Ships /command-guard:check and a user-owned allow-file the agent cannot edit.

by cjthompson

PreToolUse hook that runs Bash commands matching configurable patterns (rspec, fresheyes.sh, ...) under an idle-hang watchdog: kills them if output and CPU progress both go flat for a sustained window.

by Ivan-WG

Meta-bundle: the user-scope baseline — the plugins worth enabling once, globally, and keeping on in every repo: secret-scanning's write-time block, candor's five-clause Stop gate and dormant terse mode, code-review's fan-in review with comment discipline, git-workflow's branch finish, hindsight's friction mining, skill-router's file-aware routing, and stack-scan's report, audit and project-tier scout. Replaces always-on-suite and quality-suite (2026-09-14). Membership rule: project-agnostic, safe as a permanent fixture, adds no interruption you did not ask for — why command-guard's ask tier is not here. Uninstall with /core-suite:uninstall. 200k windows: read the README first.

by johantor

Team-style feature delivery for Claude Code: a captain agent plans the work and delegates each step to backend, frontend, test, and design specialists behind hook-enforced guardrails, with a consolidated review gate before anything ships.

by Victor Kzam

Claude Code workflow plugin: design/build/ship skills, orchestration agents, and guardrail hooks.

by Thanh Nguyen

Drive a Forge issue tracker from the terminal — browse, read, file and comment over the MCP endpoint with no MCP client connected, writing every issue in natural Vietnamese through the bundled vi-natural CLI. Also the global half of the two-level rule arrangement: it guards the shell commands that cannot be undone, stops once before a memory or skill write, and routes every code file a call wrote — through Edit or through the shell — to the project's own eslint, whose rules stay the project's.

by Devrik

Deterministic gates for Claude Code: destructive-command blocks, protected paths, delegation briefs, spec-driven flow and session-start validation. Selection lives in config, not in code.

by Dongmin Yu

Seven defensive hooks: blocks recursive rm on $HOME, download-and-execute pipelines, secrets-file access, path-less formatter runs, credentials in the staged diff, and the two zsh quoting mistakes that fail silently; warns on stale lockfiles and shellcheck findings.

by Ivan-WG

Prices every output as a debit and asks how much. Five surfaces, one bar each: code, tests, comments, files, and actions — a tool call, subagent spawn or verify pass whose result cannot change the next step is a paid no-op, and nothing else here prices that. Only four things buy more than the minimum: blast radius, an observed defect, a stated criterion, the user asking; the excess must name which, where it happens. Minimum means risk coverage, not count — never delete a test to hit a ratio.

by Anselm Hahn

Reworks a Claude Code plugin to the official Anthropic standard the way a failed part is reworked to spec: a calibrated review of its skills, agents, commands, workflow prompts, hooks, scripts, report viewers, manifest, README and the developer docs that wire it in, then a tier-gated fix pass that applies only what a haiku or sonnet remediator can do without judgement, blind-verified, under a PreToolUse hook that denies every edit outside the fix lock. Nothing is graded until the finder passes a sealed hold-out; opus- and human-tier findings are surfaced, never auto-applied. Read-only except analysis/nacharbeit/ and, during a fix pass, the files its lock names.

by David Pham

Pasture lifecycle hooks for Claude Code — inject beads workflow context on session start and before compaction, and enforce shared-worktree git discipline for worker agents. Installed default-off; enable explicitly.

by Karl

Wake your session on CI failures and merge conflicts — no foreground polling, no comment-channel injection.

by Nery Samuel Murillo

Format files edited by Codex with Prettier after each change.

by Alexandre Oliveira

Refuses the hand-edit of a governed roadmap, changelog or rationale file and names the command to call instead, runs the gate as the turn ends, and offers add, ship, pick and lint as MCP tools and as slash commands over the one engine.

by nuttaruj

Persistent memory for AI coding agents. Captures your sessions, consolidates them into a searchable markdown wiki, and hands the next session a short list of pointers. One binary, nothing resident.

by Ivan-WG

Secret-leak prevention: a PreToolUse hook that blocks a Write/Edit introducing a high-confidence secret (cloud keys, private-key blocks, provider tokens, assigned secret literals) before it reaches disk, plus a /secret-scanning:scan command to sweep a path or diff on demand. Fail-open, high-confidence-only, with a fixture escape.

by Christian Attard

Plain English harness. States the budget before the agent writes, and reads what came back.

by Ivan-WG

File-aware skill auto-routing: a PostToolUse hook injects one additionalContext directive to load the relevant best-practice skill(s) when you edit a matching file (PHP/Laravel/Blade, React/Vue/Inertia/Livewire via .tsx/.jsx/.vue, plain .ts/.js/.py/.go/.rb/.rs, SQL and migrations (one engine-agnostic rubric, plus a MariaDB row where that engine is detected), components, tests, Dockerfiles, OpenAPI), a SessionStart hook primes a repo-tailored skill index, and low-confidence content signals flush as one digest on your next prompt (SessionEnd keeps the ledger). Every relevant skill co-fires per edit; fail-open, once per skill per session.

by Claude Code Skills

Project-focused specification and task management with PLANS taxonomy, ADR-style architecture decisions, SMART acceptance criteria, SEAMS-enhanced clarification, tech-specific designer agents, anti-pattern detection (50 patterns across 5 tech stacks), 4-level verification (stub detection + wiring checks), execution orchestration (deviation rules, checkpoint taxonomy, auth gates, session continuity), specialized analysis agents, git checkpoints, mandatory approval gates, and session-memory hooks. Outputs to speckit/ directory.

by Tamir Cohen

27 bundled skills, smart worktree hooks, statusline, and MCP server stubs — plan, implement, review, debug, multi-agent setup, and audit code from one plugin.