Recently added

セッションアーカイブの意味検索と教訓蒸留による自己改善基盤. Includes 1 skill and 1 MCP server.

Real-time statusline HUD for Claude Code - context health, tool activity, agent tracking, and todo progress

Lean Codex research-state workflow with hosted MCP research state, typed artifacts, reviews, and cloud SSH sandboxes.

讯飞虚拟人交互平台全流程接入助手，支持 Web 对话模板、数字人直播、三端 SDK 开发、WebAPI 报文接入、模型配置、知识库和故障排查。

讯飞虚拟人交互平台全流程接入助手 - 从零代码快速搭建到 SDK/WebAPI 深度集成,支持 Web 对话模板、数字人直播、三端 SDK 开发、后端报文接入

Domain and DTO design, code smells, type design, errors, comments, and React. The guidance skills plus the lint rules and metric gate that enforce what tooling can.

DevOps and SRE intelligence for ChatGPT Codex — investigate incidents, search Datadog logs, diagnose CI/CD pipelines, check deployments, and operate across AWS, Kubernetes, GitHub, Jira, Sentry, Cloudflare, and more through a secure remote MCP gateway.

Make one or more Obsidian vaults Claude Code's durable, token-efficient memory: cross-platform, multi-vault setup and health checks for the Local REST API bridge (one MCP server per vault), a vault contract guard whose frontmatter and ASCII rules ship off and whose canvas shape check ships on, gardening and reflection agents, canvas and map-of-content generation, and graphify wiring into a separate codegraphs vault - no vault path hardcoded.

Local models on your own GPU via Ollama, in two halves: three MCP tools that index a repo or folder into an on-disk vector store, search it, and answer from it inside your current session; and a `localgpu` shell/proxy that launches a separate Claude Code session running entirely on a local chat model, through a proxy translating the Anthropic Messages API into Ollama's chat API. Six commands cover setup, health checks, indexing, search, ask, and an honest account of which crew roles a local 7B can and cannot take over. No hooks and no agents; nothing leaves 127.0.0.1.

Gizmoduck — run Nuclei vulnerability scans on websites and hosts, diff against previous scans, and turn findings into triaged reports and ServiceDesk Plus tickets. Runs on WSL/Linux and Windows.

A small, honest virtual dev team: context-isolated agents, deterministic gates, file-backed work state. Optional Codex QA, optional Jira or ServiceDesk Plus, optional Obsidian Kanban tickets, optional Obsidian memory.

The shared operating language for agent-built helper CLIs — one visual + interaction + safety language across interactive bash TUIs and python helper scripts, so every tool an agent builds looks, feels, and behaves the same.

Terminal-aesthetic review boards for in-browser design iteration: curated shelves and monster indexes of HTML artefacts. Explicit invocation only. Plugs into standalone design skills when installed; persistence via the adjudant plugin, with optional Cabinet voice flavour.

A review-only, sandboxed second opinion from a Gemini-family model via the Antigravity CLI (agy). Eight subcommands (review, megareview, wip, sanity, name, compare, save, harvest) with pinned Gemini models so the answer never silently comes from another model family; the reviewer is read-trusted to the project root, write-sandboxed, and proposes edits as code blocks that Claude applies. Pairs with adjudant for vault context and persisted reviews.

The Cabinet of IMD Agents — a flavour layer for Claude Code. Eight college classmates with distinct personalities, voices, disciplines, and pairings serve as specialised web-development agents. v3.0.0 sunsets all functionality (vault writes, session anchors, gate enforcement, hooks for state) and keeps only character, voice, and working disciplines. Persistence is delegated to the adjudant plugin when active.

Operate an Obsidian vault from your code project. One command, /adjudant, with seven verbs: connect onboards a project and asks where it lives; status reports where you are, what is wrong, and what is stale; clean removes what the vault does not need; dream reads the prose and reports what only judgement finds; draw builds diagrams, canvases, and bases; board runs a self-hosted kanban; orchestrate designates this session as an orchestrator. Schema-locked vault writes, cost-gated heavy verbs, and ambient hooks that keep session notes, handoffs, and the board current. Stdlib-only Python helpers, no build step. The board is a real Beans front end as of Board 2.0. Pairs with hookify.

Deployment commands for this project (GitHub Pages today, more targets like Vercel can be added as additional commands/*.md files)

Discord channel for Claude Code, thread-per-conversation. A daemon owns the gateway and guarantees delivery; each Discord thread is its own Claude Code session. Manage access via /discord-threads:access.

MeMesh — agentic memory for coding agents.

Executive AI personas (incl. Repository Steward) and the governance establish/audit/migrate skills for agentic-governance adopters

Visual output surface for Claude Code: display text, images, tables, plots, and interactive elements

Automatically titles every Claude Code session "<project> - <topic>" from the first prompt. Works in the CLI, VS Code and Remote Control, on Windows, macOS and Linux.

Writing style guides for technical communication — PR descriptions, issue comments, commit messages

Custom status line — dir, git branch/dirty state, model, context %, usage rates (robbyrussell-style)

Structured implementation plan creation with context gathering, approach exploration, and revdiff review

Shared global CLAUDE.md rules — plan-first workflow, commit hygiene, git hygiene, CLI best practices

Git workflow skills — squash-rebase onto main after a squash-merged parent branch

Collaborative design dialogue — turn ideas into designs through incremental questions and approach exploration

Bootstrap or sync the development pipeline in any repo. /super-bootstrap is a thin orchestrator: it installs/syncs the generic harness runway (CLAUDE.md, AGENTS.md foreign-executor contract, skeleton docs, path-scoped rules, the core plugin pin, default-on harness hooks (commit-channel routing raw commits to the gateway-inline commit door; consult-check pair injecting prompt-time doc recall); a plugin-owned runway-version SessionStart hook flagging a runway receipt that lags the installed plugin and routing to the re-run; monorepo tier per-package fan-out; adopt mode retiring a consumer's superseded fork skills/agents and backfilling skeleton sections added since bootstrap; opt-in earn-gated scale module: parked + test-queue + outward + venue map), seeds product GAP cards on greenfield, and gates stack-matched skill/MCP/hook curation on substantive seed docs. Bundles commit, todo, log, triage, triage-report, help, merge, drain, check-docs-consistency, and optional release-init — all namespaced under super-bootstrap: except the entry. Per-skill contracts live in each SKILL.md frontmatter (canonical catalog); plugin README carries the index. Solo dev workflow.

Capability discovery, durable tasks, Workflow and orchestration trials, multi-source grading, comparable evaluations, Signoff-gated Skill proposals, and safe Skill publishing.

A fidelity-first academic writing Skill for Chinese and English research papers.

Task-specific frontend workflows with typed product structure, capability-first implementation plans, and SHA-bound design/runtime evidence.

Connect to Affinity by Canva's local MCP server for SDK scripting and document previews.

Connect PeerMe to Claude, Codex, and ChatGPT through JoAi's hosted MCP app server.

Connect Poll to Claude, Codex, and ChatGPT through JoAi's hosted MCP app server.

Record evidence — an operation video, screenshots and a runbook — to attach to a pull or merge request.

Record evidence (video + screenshots + runbook) for a merge request

Connect ProjectX to Claude, Codex, and ChatGPT through JoAi's hosted MCP app server.

Connect Rating to Claude, Codex, and ChatGPT through JoAi's hosted MCP app server.

Portable methodology, tools, and repository context for AI coding agents.

Databricks Lakeflow, Workflows, DAB and Unity Catalog conventions and review skills.

Everyday commands and agents used across all projects.

An AI-native software organization: agents, competencies, and playbooks for map-reduce multi-agent delivery.

From a vague startup idea to a complete research-grounded founder artifact pack: grill the founder, run deep research, generate PRD/whitepaper/VC memo/validation boards/financials/visuals, then audit coverage.

SDLC automation: MCP server and Claude Code hooks for software development lifecycle workflows.

Practical AI tools from Recursive Systems.

Two-agent editorial system: an editor agent applies five named schemata, an evaluator agent independently scores the output against measurable features before seeing the editor's trace, and they loop until a reader without the author's domain expertise can follow the reasoning. Built from cross-domain translation of forensic linguistics, species counterpoint, oral-formulaic composition, and collaborative voice reconstruction.

Claude Code integration for LycheeMem structured long-term memory.

Motion forensics for the animated web. Four jobs against a live URL: decode (analyse motion + build, no code), clone (URL → React + Tailwind), verify (AE/SSIM + motion-parity score vs. an existing impl), extract (raw JSON dump). Extracts real CSS and animation params (GSAP / Framer Motion / Lenis / anime.js) directly from JS bundles, then verifies with AE/SSIM diff — no vision tokens. Three skills: ui-reverse-engineering (decode/clone/verify/extract entry points, supports Webflow IX2 / Next.js / Tailwind v4), ui-capture (reference + transitions), visual-debug (AE/SSIM diff + computed-style mismatch diagnosis).

Live activity view for AI coding sessions — know when an agent is blocked on you, and where in the repo work is happening.