Custom agents

What's your vibe tax this month? Local-only receipt for what Claude Code and Codex would cost at API prices. One command, nothing uploaded.

手绘白板风格技术讲解视频生成器——从演讲稿到成品视频的一站式 AI 工作流 skill。

Hands-on demo of two agent-friendly knowledge formats: Andrej Karpathy's llm-wiki pattern and Google's Open Knowledge Format (OKF) — the same Claude knowledge base organized both ways.

Eighteen QA workflows for Claude Code: Playwright-BDD UI, API, integration and multi-service tests, exploratory testing, bug filing, fix verification, PR test impact, flaky triage, release readiness, load tests. Works with GitLab, GitHub, Jira, Azure DevOps and Bitbucket.

Access US federal award, recipient, agency, and spending analytics data from USAspending.gov via MCP. STDIO or Streamable HTTP.

MCP server that gives AI agents the whole of GIMP 3: every PDB procedure, every GEGL filter, renders, recipes. Windows-first. mcprush.com-c9db3f4eb5d001aefc5ae9c6

Give AI agents eyes on Android and iOS apps — structured UI elements with stable IDs, not pixels. CLI + MCP.

Declarative, event-driven loop and graph orchestration for herdr — run Claude Code, Codex, opencode and pi together until the work converges

Fetch transcripts from Bilibili videos via yt-dlp (primary) or Chrome DevTools Protocol (fallback). Only works for videos with AI subtitles enabled.

Claude Code-style /goal for OpenCode v2: set a finish line and the agent keeps working until it's met.

Gemini 4 Argon Cookbook — 1M Context Continuation, Deep Reasoning Thought Signatures & Interactive 3D WebGL Studio — by Robina Mirbahar

Free offline dictation app for macOS. Push-to-talk voice input for Claude Code, Codex, and any app. Runs 100% locally on the Apple Neural Engine.

Envelope-encrypted team secrets your AI agent can use but never read. Zero dependencies, zero servers — your git repo is the backend.

ENS + SNS name resolution, *.threews.sol subdomain minting, and pay-by-name in one import. The three.ws agent-identity SDK.

Score and disclose fixed supply, authority revocation, allocation, and launch settings.

Size ZK proof bytes, public inputs, and instructions before building a v1 transaction.

Text/image/sketch to a textured, rig-ready 3D GLB in one call. The three.ws Forge generation SDK: free TRELLIS lane, pay-per-call USDC tiers over x402, and auto-rigging.

Run coding-agent CLIs on your machine and reach them from a browser as a terminal.

Catch fabricated RAG citations before they reach the user: a zero-token verbatim gate + a burden-of-proof judge, framework-agnostic.

The only AI coding app you'll need. Run agents in parallel. Hand off tasks that come back as pull requests. Draw on screenshots to show exactly what's wrong. Native macOS.

The AI-native, open-source browser your agent can drive

Blowhorn - Social media manager

Turn any spare phone into a live status display for your AI coding agents — Claude Code, Codex, Cursor. Local, zero-config.

Verified skill marketplace for AI agents. Turn proven IDE work into tested, cloneable skill repositories, publish through MCP, and earn USDC from x402-powered sales.

A programming language designed for AI agents. 76% fewer tokens than Python.

Live visualizer for Agent Substrate: watch AI agents restore from snapshots, queue, run in gVisor sandboxes, and checkpoint back to object storage

Stop AI agents from shipping bad code. Deterministic architecture, security, API, database, and UI quality control for coding agents.

AI PDF Assistant that extracts, chunks, and answers questions grounded strictly in an uploaded document — with built-in hallucination control. Week 3 of a 90-day AI Engineering roadmap.

Agent gateway for external tools — Rhumb Index scores services; Rhumb Resolve routes supported capability calls through governed execution rails.

Play Heroes of Might and Magic III (VCMI) with your AI agent - any agent that can run commands (we use Claude + Claude Code): in-game chat, live turn reports, the agent as a hotseat player, cheat-free GM commands, and a map toolkit for writing love letters into maps. 和你的 AI 一起玩英雄无敌3。

Inline charts for AI coding agents in the terminal. OpenCode and Claude Code plugins that turn chart blocks into line, candlestick, heatmap, pie, KPI tiles, drawdown and 15+ other chart types, themed and live while streaming. Zero dependencies, works over SSH.

A curated list of hooks, guides, and tools for Claude Code hooks (PreToolUse, PostToolUse, SessionStart, Notification, Stop, and more).

Transparent MCP security proxy: tool metadata integrity, call validation, and response analysis. Apache-2.0.

A GUI terminal multiplexer with remote control — native panes that run any CLI (Claude Code, Copilot, Codex, aider, a shell), your project's markdown rendered right in the pane, and approve, deny or prompt your AI sessions from your phone over Telegram. Notifications tested on macOS and Windows; Linux is unverified.

Next-gen production scheduling & multi-objective optimization for discrete manufacturing. Bridges the gap between theoretical APS and real-world shop floors by combining a easy-extensible hybrid algorithmic core with LLM workflows for dynamic scenario analysis, custom constraint injection, and ERP/MES/Excel integration. Self-hosted.

Fully automated, 100% free AI pipeline that generates and posts faceless Instagram Reels — while you sleep.

Shared memory, focused engineering practices and reliable handoffs for coding agents. Give your agent one setup prompt.

A touchscreen companion for Claude desktop (Raspberry Pi 4 + 7-inch display)

Self-hosted, Markdown-native personal context infrastructure for humans and AI agents.

Generate machine-readable agent endpoint, pricing, and capability metadata.

Translate compact artifacts between Solana memo envelopes and EVM calldata records.

Prepare an atomic Token-2022 create, mint, and revoke transaction plan.

在麦当劳 MCP 真实菜单里做套餐轮次级求解 —— 140 种组合本地枚举，只调 3 次接口。实测热量 -40%，价格经官方接口验证。

Permission firewall, auto-approve hooks & command guardrails for 24 AI coding agents — Claude Code, Codex, Cursor, OpenCode, Antigravity, Copilot, Cline, Kiro & more.

Did the agent actually test that? Gutcheck guts each changed function and reruns only its test at the done-claim—proven, hollow, or no binding test, with receipts. Stop hook · GitHub Action · CLI.

DSH sandbox tester: process-isolated testing ground for DeepSeek Harness

107 production-grade QA engineering skills for AI agents — frontend, API, performance, mobile, security, CI, and AI/LLM testing. MCP-ready: works with Claude Code, Claude Desktop, Cursor, Zed, Windsurf, and any MCP host.

The acceptance layer for bug fixes: a failing test sealed for every bug, then used to grade every PR that claims to fix it — human or AI agent.

Read-only DegreeWorks CLI that lets an AI agent build your exact KSU semester schedule — real sections, times, CRNs, verified prereqs — from your degree audit

Curated Agent Skills for AI coding agents — Claude Code, Codex CLI, Copilot CLI, Gemini CLI and OpenCode. Each skill is a self-contained SKILL.md following the Agent Skills spec.

A second brain any AI agent (Claude Code, Codex, Gemini) can operate. PARA-based, multi-agent, ready in 15 minutes.

Provenance-aware authorisation for LLM agent tool calls. Denies a call when a consequential argument came from untrusted content. Stdlib only.

A completed autonomous AI build: an agent built this month-long project on its own under a verification discipline where every commit carries cryptographic receipts of what actually ran. All ten units sealed; builds and runs on one tested macOS profile.

Run the Google Colab GPU MCP server on Termux/Android: source-built pydantic-core + maturin, DNS-over-HTTPS fix, one-command installer. Call Colab T4/L4 from opencode, Claude Code, Cursor.

YAML-defined agent coordination workflow for Herdr: roles, small/medium/large graphs, state validation

以Openclaw架构思想构建的基于Deep Agents、MCP、OpenSandbox、FastAPI 与 Vue 3 的多用户智能采购协作平台，支持多 Agent 编排、异步任务、SSE 流式交互、HITL 审批、Opensandbox沙箱隔离与 ERP 数据分析。

Kickstarter for AI Agents - Crowdfunding platform where agents fund agents

Let Claude (Code & Desktop) delegate tasks to any OpenCode model — Grok, Kimi, GLM, Qwen, DeepSeek… — while staying the orchestrator. MCP server + Claude Code plugin.

Resumable completion protocol for coding agents: obligations live in a durable task record, completion is computed by a validator, and a Stop hook routes execution back to unfinished work.

Search and retrieve researcher profiles, works, affiliations, funding, and peer review records from the ORCID registry via MCP. STDIO or Streamable HTTP.

Agent skills for Claude Code and other file-reading coding agents — 130+ SKILL.md procedures for deep research, OSINT, production agent auditing, design, animation, deployment and technical writing. Each carries templates, decision trees and documented anti-patterns.

AgentOS: Serve agents over API, MCP, and chat interfaces like Slack

Create attributed derivative-token manifests with parent mint references.

Compose independent instructions and estimate whether they can land atomically in v1.

Plan an explicit v1 loaded-accounts-data-size limit with headroom.

One abstraction, 146 SaaS connectors. Agent skills for Apideck's Unified API — integrate Salesforce, QuickBooks, BambooHR, Jira, Shopify and 140+ more by changing one string.

Quality gates for Rust: new code must be clean, and old debt has a limit that each fix lowers.

Save and restore work snapshots using a shadow Git repository

Python SDK for The Colony (thecolony.ai) — the official Python client for the AI agent internet

Evidence-backed code context for AI coding agents: semantic + lexical search, a cross-project code and contract graph, and versioned team memory across all your repositories — served over MCP, CLI and a local panel.

Cartel Radar — локальний пошук ніш на YouTube: реальні дані через yt-dlp або YouTube API, кратність до медіани, радар свіжих проривів, крива уваги аудиторії

Yandex 360 toolkit — Tracker, Wiki & Forms via CLI, MCP server, Python SDK & Claude Code plugin ✨

Isolated Azure CLI (az) profiles for PowerShell: switch tenants, accounts and subscriptions per terminal, not machine-wide. Safe for scripts, CI and parallel AI agents (Claude Code plugin).

DCC-MCP agent skills for Autodesk Maya and mGear Shifter: inspect rigging components, create guides, build character rigs, and export or import FBX/Alembic assets through typed tools.

Background computer use for Windows: coding agents (Codex, Claude Code, any MCP client) drive GUI apps on a hidden desktop with real mouse and keyboard while you keep using your PC. Each agent gets its own layer.

Open source AI employees — roles, skills, and missions as Markdown files

Self-hosted control plane for Amazon QuickSight that authors as well as administers: an assistant that plans dashboards and datasets with a wireframe before anything is written, a Studio that fixes what QuickSight refuses, account-wide playbooks, lineage, activity and the same API for people and agents. One CDK deploy into your own AWS account.

Global prompts and subagent roles for coding agents, deployed to every harness you use.

AI usage meters on a $5 AliExpress desk clock - Claude, Ollama Cloud, Antigravity. MIT fork of click6067-ship-it/token-tv; PRs welcome for more providers & faces.

Aster RPC -- peer-to-peer RPC framework with identity in the connection. Machines authenticate to machines, on behalf of users. Built on iroh QUIC + Apache Fory + capability-based credentials.

Gojiberry AI — independent third-party profile of a public API surface, by API Evangelist. Gojiberry AI is a Y Combinator-backed (P26), EU-hosted AI sales development platform that turns a company website into an autonomous go-to-market agent. The agent detects buying and social intent signals, scores prospects against an ideal customer profile, bu

Learn AI Evals with Manjeet

Nine Claude Code skills I wrote and twelve third-party ones I use: spec, product check, research, design, review, project audit.

MCP server that indexes codebases into a SQLite database, exposing file metadata, exports, dependency graphs, and per-file change tracking.

Read ERC-8004 agent trust scores and attest agent-to-agent feedback on-chain, in one import. The three.ws reputation SDK: canonical Identity + Reputation registries on Base and 14 other chains.

Define filters for new meme launches that use atomic v1 creation patterns.

Prepare encrypted-message metadata and recipient hints without exposing plaintext.

Local code review hub for AI coding agents (Claude Code, Codex, Cursor, opencode). Agents submit with one CLI, you review files ranked by importance, feedback flows back. GitHub or plain git.

An agentic canvas where the board is just files: every block is a real file, and any coding agent can work on it. Self-hosted.

Claude Code plugin to bootstrap and govern AI agent architectures (Claude Code, Cursor, Codex) — roles, task lifecycle, local auditing, and a secrets anti-leak guardrail.

MCP server for Tribeunal — 47 tools and 8 Agent Skills that put humans and AI agents on the same jury

Never lose your AI context again. Open-source memory that follows your projects across Claude Code, Codex, Gemini CLI, Cursor and every computer.

Complete tutorial for building AI assistants with LangGraph - from basic chatbots to production-ready agents with tools, memory, and streaming

Safe and auditable interface between AI agents and Anki.

공공데이터 덕후, 오데덕. AI 에이전트가 목표에 맞는 자료를 찾고, 자동 활용신청·수집·연결·분석을 거쳐 결과와 근거를 만들도록 돕는 오픈소스 실행 플랫폼.

Template repository for controlled AI-agent development cycles

Direct AI manipulation of CapCut project files via Model Context Protocol. A local tool for AI agents built with Elixir and OTP.

Spend intelligence where it matters. The inference control plane for coding agents: usage-limit warnings, model and effort advice, stuck detection and budget-aware task runs for Claude Code and Codex. 100 percent local.