Workflow plugins

A shared planning surface between you and your coding agent. One JSON file is the state; a self-contained HTML page is the view; a gate catches drift; session hooks brief the agent automatically.

1 skill · hooks

While you read a PR, Claude answers questions and runs things but queues change requests instead of editing. Turns on automatically after `gh pr create`.

1 command · hooks

Adversarial review of a proposed design against the code it lands in — seven named tests from DDD and SOLID, read as a slice across every layer. Before implementation on the spec, and again on the branch to catch drift.

1 skill · hooks

How a change ships: user stories, the design interview, tests first, build, review, proof, then the pull request on your go.

2 skills

Designs and reviews LLM prompts for education: rubric grading, student feedback, and lesson, worksheet and exam generation. Splits oversized prompts into smaller calls, checks revised prompts, and builds new ones from a rubric. Also reviews any other prompt. Covers Claude, Gemini, Gemma 4 and DeepSeek V4.

1 agent · hooks

Lane agents and workflow skills.

2 skills

Token-efficient agent context, as two skills: /context:setup and /context:audit.

2 skills

Install judge-jev into a repository or a parent folder.

1 skill · hooks

Create and run web, iOS, and Android tests, investigate failures, and author code-managed flows with Autosana.

1 skill · 1 MCP server

Generate rich, self-contained HTML documents as alternatives to markdown

1 skill

Run a project as an orchestrator driving implementation subagents against a tracked backlog: break work into items, brief agents that work in isolated worktrees, verify what comes back rather than accepting it, and merge through a path that mechanically refuses red checks. GitHub issues by default; a local store where the repo is not yours.

1 skill

Shared agent layer. Backend pack ships first.

hooks

The Spec-First Collaborative SDLC skill and its adoption templates.

1 skill

Python engineering conventions for tooling, type-driven domain modeling, validation, testing, and documentation.

1 skill

Forces git worktrees to be named after the work they hold, and refuses to silently resume an existing one.

hooks

Quizzes the developer on the code built in the current session to verify real understanding.

1 skill

Restrict the top-level thread to orchestration tools; force work into subagents.

hooks

Structured plan-then-execute workflow with per-step documentation in pipeline/.

1 skill

No-sycophancy, evidence-grounded response discipline: a UserPromptSubmit injection that states the rules each turn plus a silent, non-blocking Stop logger that records banned-phrase hits for tuning.

hooks

Orchestrate multiple coding agents across separate issues using Herdr, with worktree-isolated workers and a persistent board that survives context clears

1 skill

Skill for orchestrating work through the delegate CLI.

1 skill

OpenShift Node team assistant for development, deployment, debugging, and workflow tasks across kubelet, MCO, CRI-O, crun, conmonrs, Kueue operator, Jira, Red Hat KB/support cases, Prometheus, and platform docs.

1 skill

Working conventions, templates and real cases for AI-assisted work, kept in the repo they came from and installed anywhere.

1 skill

Hybrid retrieval, cited answers, RAG metrics, and agentic failure classification for industrial QA demos

1 skill

An engineering collaboration layer for coding agents. Install the CLI (install from the GitHub repository) and run `keelson init` once to create the lightweight project integration, discovery shims, and optional Claude hooks.

1 skill

Audit your own Claude Code context, find which MCP servers and skills cost tokens or sit unused, and disable them reversibly

1 skill

follow-through: carrying work from a vague idea through objective, plan and execution without lying to yourself about where it stands

1 skill

用于减少 LLM 编码常见失误的行为准则：编码前先思考、简洁优先、精准修改、目标驱动执行

1 skill

Author, run, and debug chainq prompt chains — multi-step LLM workflows defined in one YAML file and executed on a local CLI model (claude -p, codex).

1 skill

Scaffold a repository for Claude Code Agent Teams (security-review subagent, experimental teams flag, CLAUDE.md, spawn prompts, fish/tmux launcher).

1 skill