MCP servers and skills: search, filter by where they run, and compare.
Build durable agent harnesses so smaller models can do repeatable, verifiable work — a versioned folder with tools, guardrails, validators, and measured evidence from every run.
SkillApache-2.0★ 4Few starsNo key
An AI operating system you own: five layers, Constitution, Memory, Skills, Tools, Focus. Send this to your AI and it builds you one.
SkillCC-BY-4.0★ 3Few starsNo key
Experimental gateway for MCP tool calls: information-flow labels, a content tracker, session rules, human approval, audit. Research prototype; synthetic data; same-author evaluation.
PythonMITFew starsNo key
Guardrails for AI coding agents, enforced outside the context window so they survive context rot and compaction. Works with Claude Code, Cursor, OpenCode, Cline, Codex and Gemini CLI. Audit your setup
SkillApache-2.0★ 1Few starsNo key
OpenPortable: portable MCP memory, skills, workflow gates, and session handoffs across AI clients.
Node.jsMITFew starsNo key
A privacy-first, review-gated memory workflow for multiple AI assistants, devices, and terminal instances.
SkillMITFew starsNo key
JVM incident analysis from thread dumps, GC logs and heap histograms - deterministic rules, evidence chain on every finding, runs offline.
SkillMIT★ 1Few starsNo key
compass: a tiny MCP server that ends each turn with 2-4 ranked, copy-paste-ready next-best-prompts (and skips when there is none). Works in Claude Code, Claude Desktop, and Claude.ai web/mobile.
Node.jsMITFew starsNo key
Practical AI agent Skills (agentskills.io) for project governance, bounded coding loops, research & session memory — shaped by real QC/production process habits.
SkillMITFew starsNo key
The open, governed state layer for AI agents — trustworthy, not smarter. The model proposes; a deterministic engine decides; every change is an event. Sits under any agent framework, MCP-compatible.
SkillApache-2.0★ 1Few starsNo key
Who stands behind it
