Recently added

Shared session bus so Codex can list/read/handoff/send/spawn with Claude, Cursor, Grok, and friends without pasting context.

Local graph memory that lets Codex recall project context across sessions.

Burp Suite MCP bridge for target-centric traffic triage, marked-flow discovery, time-window search, replay, rewrite automation, selection handoff, and evidence export.

Turn deadlines, workload, and blockers into a realistic task schedule.

Persistent memory system for Codex - seamlessly preserve context across sessions

一套面向新手的 Codex 工作流系统，用于项目总控、学习引导、调试修复、漂移审计和阶段收尾。A beginner-friendly Codex workflow system with 7 collaborative skills for project management, learning, debugging, drift detection, and phase closeout. Bilingual (CN/EN).

Use Ontoly Software Graph evidence, CLI reports, MCP, and Agent Skills for deterministic TypeScript codebase understanding.

Multi-agent development workflow orchestrator with 12 specialized agents

Experimental MCP server for bitbank — botters lab community edition. Market data, technical analysis, charting, and Private API trading tools.

Read-only guidance for browsing the reviewed DS4CC catalog through its MCP tool.

VeigaPunk Xbrd Selector plugin (OpenAI + xAI Grok; no Claude).

VeigaPunk Xbrd Gdsp Fknpft plugin for Codex marketplace access.

VeigaPunk The Puppeteer plugin for Codex marketplace access.

Empirical network configuration audits with adversarial validation.

Drive Grok-only web UI features through agent-browser and an authenticated CDP profile.

Drive Kimi web UI chat through agent-browser and an authenticated CDP profile.

Drive NotebookLM-only web UI features through agent-browser and an authenticated CDP profile.

VeigaPunk Spoderman plugin for Codex marketplace access.

Sekhmet — always-available Codex Titanium swarm substrate (xbreed L3, max 64 runners).

Set Grok Bot as a Vegapunk round table. Say 'set grok bot for me' in your preferred CLI.

Curated skill inventory and workflow helpers for Codex.

Reusable command and shell routine workflows for Codex.

Curated agent profiles for development, review, research, and orchestration.

VeigaPunk Infinizoom plugin for Codex marketplace access.

Reports how much of your Codex usage limit is left as turns of work rather than a percentage, prices a job before you start it, and counts the other agents sharing the same budget.

商家营销内容助手：从公开链接或手工资料出发，制作、审核并导出内容。. Includes 4 skills and 1 MCP server.

商家营销内容助手：从公开链接或手工资料出发，制作、审核并导出内容。. Includes 4 skills and 1 MCP server.

Block-addressed docs for your agent: read or replace one section of an existing Markdown file instead of loading the whole thing. Ships the CLI, an MCP server, and a code-graph navigator. Optional: upgrade a doc to GEML for validated writes and per-block history.

Block-addressed docs for your agent: read or replace one section of an existing Markdown file instead of loading the whole thing. Ships the CLI, an MCP server, and a code-graph navigator. Optional: upgrade a doc to GEML for validated writes and per-block history.

Linear issue-comment and issue-link formatting conventions for tracker writes. Not the Linear MCP server registration — see linear@claude-plugins-official for that. This plugin carries only the linear-formatting skill. Install at project scope in any repo whose tracker MCP server is Linear.

Repeatable evidence-backed triage of this repo's open GitHub issues: batched parallel evidence-gathering, cross-batch synthesis, a claim-verification pass, and a disposition report. Report-only — executing dispositions stays a manual follow-up. Every dispatch runs general-purpose under the operator's own ambient gh credentials and unrestricted Bash, with no command-level enforcement beyond prompt-level standing rules. Issue and comment text is untrusted input from anyone who can file an issue, and run artifacts persist unredacted outside the repo. Install at project scope only where the credential's reach, that input exposure, and that retention are all acceptable.

mattpocock-skills is an open-source Claude Code plugin indexed from GitHub. Review its manifest and source before installing.

Policy and hooks that make the mcp-sqlite-memory server a second brain for long-running work: raw capture by hooks, curated entries by template, LaTeX-only math, a controlled tag vocabulary, and topic resume across months.

spaceagency's practice codex and WA project toolkit as a Claude plugin. The GROUNDCONTROL documentation codex (what information goes where in a construction set, ArchiCAD detail workflows, drafting principles, and the spA voice) plus WA workflow skills — site analysis, due diligence, planning-envelope, materials research, and presentations — reached via the /studio router and governed by the practice rules/.

在本机安全地把 PlantUML 源码渲染为 SVG 或 PNG，并可直接嵌入 Markdown 文档。

Shipd commands and skills. Everything here is invoked under the /s: namespace.

Game Atelier — 游戏项目资产工坊，本地 Web UI + AI 美术 / 游戏 UI / 视频工作流

Deploy AWS CDK apps safely with cdkd (CDK Direct). Guides AI coding agents through installing cdkd and bootstrapping, previewing, deploying, verifying, and destroying dev/test stacks with explicit safety boundaries.

GALA code intelligence for Claude Code via `gala lsp`: diagnostics after every edit, hover types, go-to-definition and references. Requires the GALA CLI (`gala`) installed and on PATH; the plugin does not include it.

Collapsed 4-entrypoint form of the code-quality-atlas suite for cloud / account-skill / context-budget installs. Lenses are bundled and loaded on demand.

A research-derived suite of 41 code-review and maintenance lenses covering correctness, readability, structure, security, performance, tests, migrations, accessibility, observability, LLM and agentic/tool-use safety, install/upgrade experience, authored-artifact conventions (e.g. SKILL.md), and repo-wide audits (architecture, dependencies, config, docs, compliance), plus a router that picks the right lenses for a given change or design doc, a pre-pass that grounds the review in the deterministic tools the repo already configures, and a synthesizer that merges their findings into one ranked verdict. Each skill is generated from a verified research atlas with provenance, refined examples, and cross-model evals.

Ultimate Frontend Skills (UFS for Claude). Build cinematic, editorial websites the way an award-winning studio does, on one route from brief to ship: a corpus of award-winning reference sites to decide against, WebGL mesh gradients, three-plane parallax with real depth, procedural and Blender-modelled three.js exploded product views, CC0 PBR materials and HDRI lighting, scroll choreography in GSAP and anime.js, and a typographic system that does not read as AI-generated. Ships a copy-in CSS chassis, motion and WebGL runtimes, a section library, a scaffolder, a headless render check, a bench detector that defers to frontend-design when it is installed, and one verify verdict.

Environment doctor for Claude Code — checks git/Node.js/VS Code/Docker/claude-mem health and suggests exact fix commands per OS

AI browser agent: browses, clicks, types, and reads real web pages from plain-English instructions.

The Data-First Engineering Protocol as an installable asset set: 16 skills, 14 agents and 16 enforcement hooks that put design before implementation and tests before code.

装进任何 Agent 的通用学习认知引擎：错题闭环、学习规划、进度诊断、原创模考与可选本地语义记忆，科目配置驱动，预置考研包。

Audit ML reproducibility with static checks and local experiment replay.

Reusable profession profiles, engineering skills, and native-agent helpers for Claude Code and Codex.

Evidence-based work-item decisions: decide what deserves implementation or filing, preserve provenance, and hand off a bounded next-action queue.

Create professional video guides with TTS narration, screen recording, and YouTube upload