All AI agent resources

Pydantic is a Python library facilitating data validation through type hints, particularly useful for AI agents, offering fast validation capabilities and compatibility with various development tools

Quorum lets multiple AI models and agent CLIs conduct structured debates using formal methods, critique, and synthesis, through a terminal UI, MCP, and a web version.

Raven is an open-source host agent that plans complex tasks as DAGs and orchestrates built-in research, coding, design, and on-call agents plus third-party agents such as Claude Code, Codex, OpenClaw, and Hermes Agent over ACP, CLI, or OpenAI-compatible APIs.

A self-hosted Telegram agent that analyzes social-media videos, returns source-linked breakdowns, and can plan or build related projects on the user's PC with command approvals.

Remembra is a self-hosted semantic memory system for AI agents, with persistent memory, retrieval, knowledge graph features, and integrations for agent frameworks and MCP.

rote compiles proven agent skills into typed, deterministic Python or TypeScript pipelines, with tests for fixed steps and typed LLM-judge signatures for work requiring judgment.

An async HTTP client for autonomous agents that handles HTTP 402 payments across multiple payment rails, enforces spend policies, recovers credentials, and emits audit traces.

A Rust research prototype for LLM runtime and agent control, with routing, memory, runtime adapters, rollback, tracing, and gated self-evolution.

Persistent, consensus-validated memory infrastructure for AI agents, with confidence scoring and natural memory decay.

MCP server for Salt, an end-to-end encrypted chat where humans and AI agents are equal contacts and can message, ask each other for approval, and pay each other. The local stdio server runs one agent identity with its own key; a hosted OAuth endpoint lets any MCP client connect a keyless agent with no setup.

SandBase Harness is a self-hosted AI agent runtime and MCP bridge for persistent sessions, sandboxed tool execution, credentials, audit logs, and replay. It supports local and container-based execution backends, whose isolation properties differ.

A hosted MCP server and agent skill for searching an index of onchain-agent tooling and checking package reports about key access, key egress, and fund actions before installation. Its install hook sends command text to Sato Hub and allows installs by default.

Sayna is a self-hosted real-time voice server that provides speech-to-text and text-to-speech through WebSocket and REST APIs, with provider integrations, LiveKit streaming, and SIP telephony.

Screenpipe captures screen and audio locally, then exposes searchable work history and agent context through an MCP server. It supports scheduled agent workflows and per-pipe access policies, with processing options that include local and configured remote models.

Self Auditing Agent pairs audit claims with rerunnable commands and includes gatecheck, a mutation tester for quality gates available as a GitHub Action and pre-commit hook.

Serena is a coding-agent toolkit exposed as an MCP server, providing semantic code retrieval and editing through language-server-powered tools.

An agent governance platform that intercepts tool calls for policy checks, human approval, and auditable traces.

since-cutoff is an open-source Python tool that shows a coding agent which public APIs of a project's pinned dependencies changed after its model's training cutoff and writes short notes into AGENTS.md or CLAUDE.md.