Browse agent skills

A local, in-page coding agent: an OpenCode server in Docker, a sidecar that drives it, and a bar any page can load.

Accessibility-first macOS Computer Use capability for DeepSeek Harness with stale-observation protection, app leases, confirmations, screenshots, and Web diagnostics

Compact tool call rendering, diff visualization, Shiki code blocks, codemode diffs, and output truncation for the Pi coding agent. Successor to pi-tool-display.

Feature-file AST for Kumiko: parse, patch and render feature definitions, the pattern library, and the codemods kumiko-upgrade runs.

Lean wrapper of pi-deep-research: one research tool (op search | extract | checkpoint | help) with trimmed startup tokens.

Graph-first reverse engineering for legacy systems (Reversa fork with graphify). Codex agents query a graphify knowledge graph of the code before reading files, so analysis runs faster with far fewer tokens. graphify is installed automatically on first ru

Zero-residue multi-agent concurrency hang watchdog & safe guard for Pi Coding Agent

channel-github: mentions in a pull request's conversation as work items, replies and approval cards, behind the factory's channel port, on Vercel Chat SDK.

channel-slack: Slack mentions as work items, replies and approval cards with buttons, behind the factory's channel port, on Vercel Chat SDK.

One shared graph for every AI assistant and device — memory and tasks, local-first, signed, auditable, direct P2P sync; MCP tools, no hosted account. | 一份记忆，每个助手，每台设备：本地优先、可审计的跨设备记忆与任务协作；提供 MCP 工具，无需托管账号。

Async Python client for the Synology Router Manager (SRM) Web API - read-only, typed, for mesh and standalone setups

Turn any LLM into a professional AI filmmaker. One Agent Skill for Claude, ChatGPT & Gemini that thinks like a director, DP, editor and prompt engineer, from brief to Seedance 2.5 block prompts. Built from a production-tested workflow.

Local procedural music editing workbench with a classic desktop interface and shared projects, MIDI exchange and selection-scoped multi-agent assistance

Tamidi payment service client, SePay payout account linking and VAT e-invoicing (VIN-CA, MISA)

Local-first, embeddable whiteboard & diagramming SDK for React and Next.js

Give Claude ears: rate AI voices & TTS (1-5 naturalness), check the script, pick the best take, and find clicks, clipping and pitch glitches in songs. 소리 듣기 플러그인

CSV 批量切分 / 去重 / 字段校验 / 清洗 / 合并 / 转 Excel（自动识别编码分隔符）

A community-driven Svelte 5 library and starter kit for ElevenLabs Conversational AI and low-latency audio streaming.

Raftly: idea to production with the judgment of a senior product and engineering org. A Claude Code plugin: /raftly

AI vision PDF translation and LaTeX reconstruction for DeepSeek Harness

Engineering skill pack for AI-assisted delivery: 183 skills, 39 roles, 25 workflows, 60 data contracts, with a CLI to install skills into your agent platform.

图片批量压缩 / 改尺寸 / 加水印 / 转格式（Pillow）

Semtindeki sitesi olmayan esnafı bul, satış şansına göre puanla, her birine kendi adıyla site kur. Ücretsiz, Türkçe, Claude Code skill'i dahil. yapaytu.com

Bidirectional bridge between DeepSeek Harness and the Neeko desktop shell: WebSocket server, health endpoint, protocol translation, and two-way session transport

Typed SDK for the Debate AI API, generated from debate-openapi.yml with Hey API and powered by grab-url (caching, retries, rate limiting, and dedupe on every call).

Worker-isolated local document text extraction without application or model dependencies.

n8n nodes for eVerbary: order certified translations for USCIS, courts and universities, follow them and get the finished files back.

Framework-free engine of the OGE UI Spreadsheet, shared by its Angular and React render layers: a dependency-free xlsx reader (styles, merges, links, notes, pictures, charts, conditional formats, opaque passthrough, import issues), Excel number formats, t

SEO skills for Codex, Claude Code, and Cursor: technical SEO audits, structured data, indexing, content planning, and implementation with seo-engineering.

Convert text into compact logical form for token-efficient RAG