Workflow plugins

One-way Telegram notifications: let the agent reach your phone when you are away from the terminal

2 skills · 1 MCP server

Tiered subagent chain: a cheap scout maps the code, a planner writes the plan, a critic attacks it, sonnet implements the well-specified tasks.

1 command · 2 agents

Project lifecycle skills: retire-project, next-free-port, deploy-cloudflare (Tier-3 gated).

3 skills

Git branch workflow skills: isolated worktree creation (worktree) and safe feature-branch sync with main (sync-main).

3 skills

A charter grants powers and limits them in the same document. Turns a two-minute interview into a project's working agreement and its enforced permission boundaries, then stays out of the way.

3 skills

Initialise, adopt, maintain, audit, or hand over a vendor-neutral workspace: the repository's files, not any provider's chat history or hidden memory, are the durable record of an AI agent's work, so the same subject resumes from another machine or another harness. Memory Bank context files, MADR decision records, a Markdown or Beads task backend declared in workspace.yaml, session records and approval packets. Ships the standard, a template, scripts to create, validate and migrate a workspace, and Claude Code hooks that enforce the session record.

1 skill · 1 command · hooks

Skills for trunk-based development: atomic commits, small PRs and a main branch that's always releasable. commit: write Conventional Commits 1.0.0 messages that respect the repo's commitlint/commitizen config. changelog: turn the commits since the last release into CHANGELOG.md entries. pr: write a short PR description with a Conventional Commits title and open or update the PR with gh.

3 skills

Consolidates Claude Code native auto memory. A SessionStart hook decides when a pass is due and has the main session start a visible dreaming:dreamer subagent, which writes a proposal and a report to a staging directory. The dreaming:dream skill shows the diff, snapshots memory in git and applies it only on approval, and a PreToolUse hook confines the subagent to staging. Requires jq.

1 skill · 1 agent · hooks

PR description generator skill for personal use

2 commands

Long-running autonomous feature PMs that keep working while you sleep

1 skill · 1 agent

Comprueba contra el repo lo que un texto afirma sobre el repo: ramas, pull requests, rutas de fichero y commits. Sin configuración.

1 skill · hooks

A beginner-safe workflow guardrail for Claude Code.

1 skill · hooks

Bans reflexive agreement and unverified completion claims. Ships the rules as a skill and a Stop hook that blocks a success claim when files were edited and nothing was run.

1 skill · hooks

Per-model reliability harness: makes the model you run (Opus/Sonnet/DeepSeek) cite every figure, abstain instead of inventing, and cross-check with a second model. Ships a skill + an optional deterministic PreToolUse gate. stdlib-only, non-blocking.

1 skill · hooks

Truth-first reporting, evidence-based claims, commit gate, documentation standards, and systematic mistake handling.

2 skills

Kanban board manager and story architect. Story-per-file source of truth with Obsidian Kanban views, Mermaid DAGs, and parallelizable dispatch wave planning.

1 skill · 1 agent

End-to-end research lifecycle: explore, foundation, design, eval, experiment, write, review, rebuttal — with adversarial pre-mortem gates at every phase transition.

2 skills

One source of truth for every AI coding agent. Writes AGENTS.md, generates pointer-only adapters, and audits for duplication and drift.

2 skills

PM 표준 기반 착수보고서 검토 스킬. 완성도·모호성·일관성·리스크를 분석하고 개선 권고사항을 제시한다.

2 skills

The /fuse and /pr-evidence workflows plus a fail-open push gate. Divergent work (scout, plan, review) runs blind and in parallel so agreement is evidence, not an echo; convergent work (build) runs coordinated against a contract the architect freezes up front. Model-agnostic: roles are slots, not model names.

1 skill · hooks

Stabilizes MoE expert routing via role anchors, consistent vocabulary, single-domain turns, and structured prompts, then enforces impact-mapping and per-change verification. For any Mixture-of-Experts model (Qwen, GLM, DeepSeek, Mixtral, local MoE).

2 skills

Review a plan and edit a draft in nvim: plan mode opens in a tmux window instead of a browser, and a draft headed anywhere outside the conversation is handed over the same way

1 skill · hooks

Repository-based handoffs so any coding agent can resume work another one stopped. After each finished task it writes the owner a plain-language explainer of what changed and why, then asks them about it.

2 skills

Analyze and rewrite CVs section-by-section against a target job description to maximize interview callback rates while maintaining ATS compliance and natural human tone

1 skill · 1 command

Arbeitet konzeptentwicklunge aus, indem ein Grobkonzept gegen die Inhalte der Kundenwebseite geprueft und die Luecken sichtbar gemacht werden.

2 skills

为跨多次会话的长期任务在全局目录 ~/.claude/project/<slug>/ 建立可恢复、可审计的档案，按 Workdir 隔离，一个会话只加载一个项目（charter/plan/state/journal/decisions），新会话冷启动时用 5 行汇报恢复上下文。Use when a task spans multiple sessions and its goal, plan, state and decisions must survive between them.

1 skill · hooks

Log the small frictions you hit while working to PAPERCUTS.jsonl, in the moment, so they collect into a map of where the repo needs sanding down. A SessionStart primer prompts the logging; /papercut forks the conversation to mine it for papercuts all at once.

1 command · hooks

Front-door skills for the Long-Horizon Workflow Framework: /longhorizon (four-condition reality interview + honest routing) and /longhorizon-strategize (design the instantiation for a specific task in any domain).

2 skills

PR preparation, branch management, and development workflow automation

1 skill · 1 agent

Alumni-first LinkedIn networking via browser automation + bulk Bright Data enrichment. Writes to your tracker via Supabase MCP. Companion to linkedintracker.app.

2 skills