By abpai
Claude-native planner/generator/evaluator harness for long-running engineering work, with optional Codex critique at high-leverage checkpoints.
Execute the approved Pi brief. Runs the generator loop, optional simplification, evaluation, and focused repair passes until the build clears the bar or repair budget is exhausted.
Turn a project request into a brief, rubric, and ordered task slices. Interactive planner phase for long-running engineering work.
Run final Pi review. Executes the full verification suite, performs holistic evaluation, and presents the scorecard with any remaining repair guidance.
Run Codex CLI for a second-provider technical opinion when a design choice is ambiguous, high-risk, or recent enough to merit a tie-break.
Run Codex CLI as an independent critic for the plan or the latest diff. Use when an outside read could materially improve confidence.
Grade a completed build or repair pass against the Pi rubric. Runs functional verification, optionally incorporates Codex review, and returns narrow repair guidance when the build misses the bar.
Execute the Pi brief as one coherent generation pass, then handle targeted repair passes if evaluation finds gaps. Use during /pi:execute.
Turn a request into a working brief, rubric, and ordered task slices. Use during /pi:plan for large or ambiguous engineering work.
Uses power tools
Uses Bash, Write, or Edit tools
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimBased on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
A collection of reusable AI workflow plugins and skills for structured planning, cross-domain thinking, multi-model tooling, and developer productivity.
This repository now ships metadata for both runtimes:
.claude-plugin/plugin.json plus the root Claude marketplace.codex-plugin/plugin.json plus a repo-scoped Codex marketplace at .agents/plugins/marketplace.json# Add the marketplace (once)
/plugin marketplace add abpai/skills
# Install planning-oriented skills
/plugin install distill@abpai-skills
/plugin install lateral-thinking@abpai-skills
/plugin install codex-exec@abpai-skills
/plugin install pi@abpai-skills
Codex currently documents repo-scoped and personal local marketplaces. This
repo includes a repo marketplace at .agents/plugins/marketplace.json that
points at the top-level plugin folders with local source.path entries.
git clone https://github.com/abpai/skills.git
cd skills
codex
Then start or restart Codex in this repo, open the plugin directory, and install the plugins from the repo marketplace exposed by this checkout.
Codex public plugin-directory publishing is still documented as coming soon, so the Codex path in this repo is repo-local rather than the Claude-style remote marketplace flow above.
pi is intentionally excluded from the Codex marketplace in this repo. It is a
Claude-native plugin that shells out to the codex CLI for second-provider
research and review, rather than being installed into Codex itself.
| Plugin | What it does | Standalone? |
|---|---|---|
| distill | Decompose complex systems into essential primitives. Codebases, papers, transcripts. | Yes |
| lateral-thinking | Cross-domain hypothesis generation. Find transferable mechanisms from distant fields. | Yes |
| codex-exec | Delegate prompts to OpenAI Codex CLI for second opinions and adversarial review. | Yes |
| pi | Claude-native planner/generator/evaluator harness for long-running engineering work, with optional Codex critique at high-leverage checkpoints. | Claude-only |
| Plugin | What it does |
|---|---|
| code-simplifier | Simplify and refine code for clarity, consistency, and maintainability |
| dead-code-eliminator | Audit for unreachable functions, unused imports, orphaned classes, stale flags |
| review-and-commit | Review uncommitted changes, then prepare safe atomic commits |
| Plugin | What it does |
|---|---|
| debate | Structured architecture debate: Claude proposes, Codex critiques, Claude synthesizes |
| cli-design-expert | Design or review CLIs for usability: flags, exit codes, TTY behavior |
| project-memory | Always-on memory via .agents/LEARNINGS.md — mistakes, patterns, preferences |
| scratch | Understand a project's internals through runnable .scratch/ exploration scripts |
| socratic-code-owner | Quiz the developer on AI-built code to ensure understanding |
| Plugin | What it does |
|---|---|
| agent-browser | Browser automation: navigate, fill forms, click, screenshot, extract data |
| beautiful-mermaid | Render Mermaid diagrams as SVG and PNG |
| claude | Run Claude Code CLI for delegation, session continuation, machine-readable output |
| try | Evaluate a new library, tool, or repo before adopting it — prompt-driven demos |
| visualize | Generate self-contained HTML visualizations for systems, plans, or code flows |
| Plugin | What it does |
|---|---|
| bun-expert | Expert Bun runtime guidance: setup, servers, APIs, testing, Node.js migration |
| dokploy | Operate Dokploy via CLI: projects, environments, apps, databases |
| Plugin | What it does |
|---|---|
| human-writer | Edit prose to sound natural and human-written — deslop model-generated text |
abpai/skills/
├── .agents/plugins/
│ └── marketplace.json ← Codex repo marketplace
├── .claude-plugin/
│ └── marketplace.json ← Claude marketplace catalog
├── <plugin>/ ← most plugins ship both runtimes
│ ├── .claude-plugin/plugin.json
│ ├── .codex-plugin/plugin.json ← optional when a plugin is installable in Codex
│ └── skills/<plugin>/
│ ├── SKILL.md
│ └── references/ (if any)
├── pi/ ← intentional Claude-only exception
│ ├── .claude-plugin/plugin.json
│ ├── agents/
│ ├── commands/
│ └── skills/pi-protocol/
└── README.md
Within each plugin folder, only plugin.json belongs inside
.claude-plugin/. skills/, agents/, commands/, and hooks/ stay at the
plugin root.
npx claudepluginhub abpai/skills --plugin piBrowser automation CLI for AI agents. Navigate pages, fill forms, click buttons, take screenshots, extract data, test web apps.
Understand an existing project by playing with its real modules in a .scratch/ workspace.
Edit prose so it sounds natural, specific, and human-written. Use when a draft feels generic, over-polished, or model-generated.
Always-on project memory workflow. Read and update .agents/LEARNINGS.md with mistakes, corrections, patterns, and preferences.
Design or review command-line interfaces for usability and automation: help text, args/flags, stdout vs stderr, exit codes, TTY behavior, config precedence.
Persistent file-based planning for AI coding agents. Crash-proof markdown plans (task_plan.md, findings.md, progress.md) that survive context loss and /clear, with an opt-in completion gate and multi-agent shared state. Manus-style. Works with Claude Code, Codex CLI, Cursor, Kiro, OpenCode and 60+ agents via the SKILL.md standard. Includes Arabic, German, Spanish, and Chinese (Simplified and Traditional).
v9.44.1 — Patch release for Gemini environment/version detection and qwen auth gating. Run /octo:setup.
Harness-native ECC operator layer - 67 agents, 271 skills, 92 legacy command shims, reusable hooks, rules, selective install profiles, and production-ready workflows for Claude Code, Codex, OpenCode, Cursor, and related agent harnesses
Superpowers Plus core skills library for Claude Code: planning, execution routing, TDD, debugging, and collaboration workflows
Unity Development Toolkit - Expert agents for scripting/refactoring/optimization, script templates, and Agent Skills for Unity C# development
Claude harness - A harness for solo developers (Vibecoders) to handle full-cycle contract development.