Engineering plugins

Ibe Dwi's agent skills.

4 skills

Find and explain code, coach code navigation, inspect and decompile Java jars, and set up free code-intelligence tooling.

4 skills

Edit, rebuild, validate and test Warcraft III custom maps with reusable agent skills and WC3 Tester game-control tools.

3 skills · 1 MCP server

General-purpose agent skills for code review and naming: a branch reviewer, a second-opinion reviewer for other harnesses, and a domain-term glossary check.

3 skills · 1 agent

Practical engineering skills for day-to-day work: PR review handling, refactoring aids, and other tooling.

3 skills

Ticket-patrolling agent — walks rounds over the frontier: claims agent-ready tickets, dispatches implementers, babysits CI, applies the ready gate.

3 skills

实用开发模块集合，提供 podspec 整理、changelog 生成和 CocoaPods 发布流程。

3 skills

Scan dependencies for CVEs, get evidence-backed upgrade targets, and measure locally whether the vulnerable code is actually reachable. Includes the Phoenix MCP server.

2 skills · 1 MCP server

Développe une User Story au périmètre strict (GitHub / Azure DevOps / fichier .md / markdown collé) avec traçabilité critère d'acceptation → code → test, gates de clarification et de validation, et rapport final structuré

1 command · 2 agents

Go skills: Rob Pike's proverbs, production-grade quality checklist (with deep reference), and version-aware modern Go syntax.

3 skills

Delegate to named Pi subagents (.md persona + model + tools, incl. MCP) and bounded coding tasks from Claude Code. Bundles a local stdio MCP server that launches Pi (pi --mode rpc), isolates code edits in git worktrees, and returns compact summaries plus diff paths — never applying changes automatically.

1 skill · 1 agent · 1 MCP server

Async workers; Controller stays free.

2 skills · hooks

Temporary, privacy-preserving trace of Codex plugin Hook delivery.

2 skills · hooks

Delegate code review and implementation work to Claude Code from inside Codex or Antigravity. Use cc when you want an independent second opinion on a diff, an adversarial pass over a risky change before merge, or to hand a substantial task to another agent. Reviews run read-only against the real repository.

3 skills

Build, test, package, and diagnose IntelliJ Platform plugins.

3 skills

Agent skills for real engineering and game-development work: a Next.js App Router performance audit method, a GitHub review-queue workflow, and a game-dialogue writing method.

3 skills

Architectural doctrine enforcement through composable lenses. Covenant for planning, Inquisition for auditing, Scriptorium for writing new doctrine.

3 skills

Opinionated skills for raising PRs end to end, driving the Codex CLI from Claude and running UI polish passes.

3 skills

Independently report Codex token usage locally, optionally share sanitized totals, and run requested Claude Code tasks with an Anthropic API key.

3 skills

Matthieu Blanc's curated agentic skills for software engineering and productivity.

3 skills

OpenLabs (Parallight Lab) — agent-oriented learning plugin for Codex CLI. Learn to build AI agents by directing them, guided by a resident mentor for the lab you are in; draw what you want to learn as a teachboard board. Zero API keys (LLM runs through Parallight's backend). Trigger when the user's message starts with :openlabs, :lab (:lab, :lab-login, :lab-start <lab-id>, :lab-status, :lab-kb, :lab-logout, :lab-exit) or :teachboard, or asks in natural language to log into Parallight, browse/start/continue a lab, check lab progress, exit a lab, draw a topic on the board, or connect the board account.

1 skill · 1 MCP server · hooks

Find out why a web app's navigation, pages, links or Back button feel wrong, decide whether the structure should change, and prove it in a real browser.

3 skills

Agent skills for software engineering, starting with idiomatic use of the libraries and frameworks a project depends on.

3 skills

A library of Claude Code skills.

2 skills

Give a repo the structure parallel AI coding agents need to work in it safely.

2 skills

Convene a council of four reviewer cats (security, deps/supply-chain, correctness/concurrency, style) over a PR or working diff, pulling in external bot reviews too.

1 skill · 1 command

Translate between generative code and the geometry it produces, keeping output assets lean instead of bloated.

1 skill · 1 MCP server

Predict where a render, simulation or training job will bottleneck, and what to change first.

1 skill · 1 MCP server

Session retrospective — links the errors, reworks and ambiguities of a session to the skills, rules, CLAUDE.md, memory, hooks and settings behind them, and applies only explicitly authorized, reviewed fixes to the project's own files

1 skill · hooks

Ship production LLM apps: eval engineering (tracing, RAG metrics, LLM-as-judge, guardrails) and deployment (containers, cloud rollout, vector-store tuning).

2 skills