Local outcome-loop + reward layer for coding agents — remember what worked, ship the next move, measure how it landed. Docs at nimroboai.com

Starter kit for building an agent harness, the map, guardrails, and feedback loops that let AI agents ship software reliably.

Reliability gateway for AI tool output: schema-stable, secret-safe, pagination-complete JSON for MCP and CLI agents.

Universal AI behavior rule framework — pin your 5-10 rules so AI doesn't drift in long tasks. Dev-scenario preset built in; switch to any scenario in one line. Claude / Codex / Cursor / Hermes. Zero LLM, zero network, zero deps.

Build system for AI coding-agent context. Generate deterministic, explainable context artifacts for Claude, Codex, Cursor, and MCP.

SOUL.md governance framework for Hermes Agent — structured memory, skill management, and operational rules

Local-first memory and dreaming automation for agents.

A decision engine for Claude Code. MIT licensed.

Instant web hosting for AI agents. One curl, one CLI, *.wooven.dev URL in <1s.

Mission orchestrator and lab for local AI. Launch a mission, watch the graph come alive — every seat runs your own models off the meter, or a hosted endpoint.

Node.js TypeScript CLI for discovering, staging, activating, and wiring reusable AI-agent assets across VS Code/Copilot, OpenCode, Cursor, Zed, Claude Code, and Pi.

Live-updating index of LLM agent frameworks shipping on GitHub, refreshed every 15 minutes

Package manager and registry for Claude skill packages - install, validate, score, and publish .skill packages with Claude Code integration via MCP

Portable, inspectable knowledge bundles for LLM agents.

A Claude Code plugin that gives Claude a self-introspection tool for its own subscription quota — so it can warn you before a long task runs out of budget.

Public Plugin Marketplace

Schema-validated JSON blocks for AI-generated reports — Markdown source, safe runtime, zero AI-generated HTML/CSS/JS. Works with Claude Code, Codex, any agent.

Beautiful, informative status line for Claude Code. Zero dependencies, cross-platform. pip install claude-status

AIHouse — 监控你的 AI Agent 在干什么。开源桌面应用，支持 Claude Code / Cursor / Codex CLI / OpenCode / Hermes。

Graph-driven reasoning over verified evidence. A formal workflow protocol for AI coding agents: deterministic Definition of Ready, falsifiable assumption gates, atomic done-transitions, and a single auditable lock file as source of truth.