Browse agent skills

Start and stop headless Claude Code sessions across a folder of projects, from your phone.

Unofficial Cursor managed inference provider for OpenCode over InferenceService/RunInference.

Local converter between LLM client protocols and named upstream providers

A fail-closed approval gate for Pi that classifies every covered tool call into six risk levels with an isolated AI reviewer and enforces per-level allow/ask/deny from local configuration.

Send push notifications to your phone from any web app or script — and ask questions that wait for a yes.

Overlay a tiered planning workflow — roadmap, plans, phase ledgers, verification gates — onto an existing repository, for coding agents.

Hide DSH read, grep, and glob tools and their prompt guidance so agents use bash instead

DSH Web GUI 审批语音提示：需要审批/回答的弹窗出现时，播放提示音并用语音播报，避免漏看

Guided, evidence-anchored reviews of agent-written code. A coding agent authors the review; you read, ask, comment and decide in the browser.

Agent skills and deterministic tools for Uniswap and Aerodrome liquidity provision.

n8n node to look up IPs, domains, and URLs on isMalicious

EN: models.dev parameter assistant for the official DeepSeek Harness Models page: per-provider one-click context / max-output / reasoning-effort fill, written in the official reasoningEfforts shape. ZH: 官方模型设置页的 models.dev 参数助手：按 provider 一键补全上下文、max 输出与推

Official TypeScript SDK for the ScamAI detection platform. One detect() for images, video and audio, plus account and usage.

Official TypeScript/JavaScript SDK for Banzami — instant Kwanza payments on the BANZA protocol

The Stratus Agent web.search contract: one tool shape, one option meaning, and one result envelope that every search backend obeys

Puppeteer integration for NestJS 12 with a Cloudflare Browser Rendering-style API

The animated diagram editor AI agents can draw on. Animated technical diagrams (architecture, sequence, BPMN, cloud icons) built by you and your agent together on the same live canvas, via WebMCP. Open-source, local-first.

Production-grade software engineering lifecycle skills for coding agents.

Installer for the psadt-deploy Claude Code skill: build, test and deploy PSADT v4.x Intune Win32 packages.

SamlPy: Semantic Assembly Modeling Language for Python - Deterministic, Zero-Coordinate, LLM-Native CAD Engine built on OpenCASCADE

Scrape npm metadata, versions and download stats for AI/LLM SDK packages — openai, langchain, @anthropic-ai/sdk, ollama — via the official npm Registry JSON API. No proxy, no auth. Returns version, license, dependency count, downloads-last-month, parse_co

A Playwright-powered testing harness for browser extensions

A Playwright fixture that turns `retrace run`'s environment into checkpoints and flow-part markers. Builds on [`@caribou-crew/retrace-js`](../js) for the env handshake and `group`/`endGroup`.

Heuristic advisory risk classifier for shell commands, with a forkable rule table, for deciding when to prompt a human.

Run a Relay session against a repository on your own machine — everyone with the link watches the agent work, one person drives.

SolidForge — loop-engineering skills for the Pi coding agent: converge → specify → implement pipeline + outcome-axis verification layers. Pi port of the Claude Code plugin.

LibreChat multi-model AI chat — MongoDB (provisioned externally), Redis sessions, optional Meilisearch, OIDC SSO

Cron-driven autonomous task runs for the dsh harness: fire a calendar cron rule and spawn a fresh agent run via ctx.agents.create(), unlike the official @deepseek-ai/dsh-schedule which only messages the live session.

AnySearch-backed WebSearchProvider for DeepSeek Harness (dsh): register AnySearch's /v1/search, switch explicitly between the official DeepSeek search endpoint and AnySearch from the Plugins settings page card, compatible with dsh v0.1.2-alpha.1 ~ rc.1 (s

UniVerscan - Universal Playwright Web Compliance Scanner