PyPI agents

Declarative agent registry — catalog, discover, and diff agent capabilities across environments

Peer AI agents coordinate through markdown messages in shared channel folders — no orchestrator, autonomous zero-token waiting, cross-platform.

Local MCP server that saves work-in-progress checkpoints so any AI agent can resume a multi-step plan exactly where a previous session was cut off.

Chief — the chief of staff for your agents and information sources

Record-and-replay for agent decision graphs: reproduce a prod agent failure as a committed regression test — and re-run your fix without live LLM calls.

Deterministic safety gate for AI coding agents

Extracts structured, token-efficient context bundles from Python codebases for AI coding agents.

Agent-focused semantic mapping and language services for mixed-language codebases.

Python SDK for Agent Communication Hub — multi-agent messaging, task scheduling, shared memory

Private, self-hosted coordination control plane for coding agents

DefenseClaw evaluators for agent-control

Agent Control Specification: a stateless policy decision runtime implementing the agent-hooks interceptor contract.

Brief framework — structured-composition pattern for agent_core

Core infrastructure for AI agents - memory, knowledge compilation, and tooling

Stdio MCP proxy — bridges Claude Code to the agent-core daemon bus tool surface with per-request backend sessions so a daemon bounce never strands the session.

Stdio MCP channel relay — bridges agent-core daemon notifications to Claude Code's wake mechanism.

KeePass-backed credential vault for agent-core agents

Discord bot adapter for the agent-core bus — one bot per agent (1:1).

Bootstrap system for hatching new agent-core beings

Inbound notifications router — deny-by-default external-event surface for agent-core beings.

Desktop notification MCP server for agent-core agents

Release-validation scenario runner for the agent_core daemon (pytest-based)

Voice synthesis endpoint for agent_core — per-agent Qwen3-TTS via ICL voice cloning

Webcam endpoint for agent_core — agent-driven on-demand frame capture

A corporation of character AI agents running on local LLMs — each agent with its own persona and app set. Name reservation; developed privately.

Per-stage token/cost attribution and silent-degradation detection for multi-agent workflow runs (stdlib-only).

Agent Crossbar — delegate review, advice, text, and dev work to local coding agents through MCP

Blame analysis for multi-agent runs: read an OTEL trace, name where quality broke

Two-tier detection/scoring/blame pipeline over Redis Streams (build spec section 4)

Detect AI coding agents from their execution environment

Flags risky changes to AI agent configs (prompts, tool permissions, model choice) in a diff.

An open-source Python SDK for building executable learning experiences.

Mine recurring feedback from coding-agent sessions and visualize it.

Proof-carrying workflows for coding agents: bind tests to exact code and detect stale evidence.

A framework-agnostic evaluation harness for tool-using AI agents. Define scenario packs, run agents consistently, score outcomes and trajectories, and gate releases with evidence.

CLI client for the Autonomous Agent Execution Engine

git blame for AI runs

Offline, lazy, agent-oriented exploration of heterogeneous folders

Reserved package name; project under development.

Typed framework for declaring and running agentic systems

Azure Cosmos DB Agent Memory Toolkit integration for Microsoft Agent Framework - semantic memory with fact extraction and user profiles.

A2A conversion helpers for app-owned Agent Framework hosting.

Agent and workflow MCP tool adapters for app-owned hosting.

OpenAI Responses-shaped helpers for agent-framework-hosting.

Telegram Bot API-shaped helpers for agent-framework-hosting.

Generic terminal client for an online agent gateway

User-space runtime policy guard with deterministic control decisions and forensic evidence.

Public Preview — Unified installer and runtime policy enforcement for the Agent Governance Toolkit

Embedded multi-tenant Agent governance engine - cost attribution, declarative policies, and audit

Lightweight recovery layer for multi-agent handoffs

A runtime environment that coordinates context, memory, prompts, tools, guardrails, output formatting, and a bounded reasoning loop around an LLM.

Framework-neutral agent control contract: interception points, agent context, verdict, and conformance test kit. Python wrapper over the canonical Rust core. Cooperative contract, not a security boundary: the host is fully trusted (see SECURITY.md).

A SQLite mailbox for local LLM agents — one HTTP API, an ActivityStreams messaging model, and a local stdio MCP server so AI coding agents can message each other. No external services.

Python client for agent-input-sanitizer: a thin bridge to the Node.js CLI for byte-identical sanitization verdicts.

Pay-per-request access to any service for AI agents, over the open x402 protocol. Declare an intent, discover a provider, and settle on-chain with a single signature.

Pay-per-request access to any service for AI agents, over the open x402 protocol. Declare an intent, discover a provider, and settle on-chain with a single signature.

Conversational memory as a knowledge graph

A project-local beginner-facing harness for Claude Code

Reliable coding-agent commands across Unicode paths, shell encodings, Git output, Python launchers, and mixed-encoding files.

Post-hoc log-analysis dashboard for AI agent transcripts (Claude Code JSONL). Reads session logs to spot order-prompt / tool-call mismatches; a log-reading tool for a human maintainer, not an anomaly monitor.

Local SQLite long-term memory for AI assistants, served over MCP

Local runtime guard for coding agents with loop detection and explainable policy decisions.

Portable, local-first self-check runtime for humans, agents, and agentic apps.

Scout LSP MCP — FastMCP + Rust/PyO3 (tasks, sessions, worktrees, blast_radius)

Turn any coding agent into a tree-searching agent — an MCTS test-time search harness for Claude Code, Codex CLI, Kimi CLI, and more.

Local-first MCP memory backend and governance console for coding agents

Two-stage retrieval + gating long-term memory engine for coding agents.

Agent Memory File System - a local-first memory store for AI agents, with semantic search.

An append-only, auditable ledger for LLM agent memory.

Local-first, MCP-native memory engine for AI agents with RAM cache and context-budgeted recall.

A local zero-token agent status API with OpenAI and Anthropic compatible streaming

Unified pgvector-backed CLI for agent notes (breadcrumbs, memories, reflections) shared across the sf2 constellation

AI agent observability with deterministic record/replay — capture any LangGraph or OpenAI Agents SDK run, replay it offline at zero API cost.

Pre-configured Haystack Agents

Safety-first synthetic data generation agent

Ephemeral, end-to-end encrypted chat rooms for autonomous agents

Python SDK for Agent Passport - AI Agent compliance proof system for EU AI Act Art.50

Offline, agent-oriented PDF exploration workspaces

Framework-agnostic completion-lie detector for AI agents -- catches an agent turn whose final claim contradicts its own tool evidence.

Give an agent the channels it needs, and only those. Install per channel, one output shape, with caching and a token budget.

AI Agent 红队安全测试平台 — 从 CLI 到 Dashboard

Statistical regression testing for LLM agents. Get a p-value on whether behavior actually shifted between versions, not just whether one run looked different. Apache 2.0, self-hostable, no SaaS dependency -- a Promptfoo alternative for the statistical testing gap threshold-based eval tools don't cover.

Local-first policy and trajectory-regression harness for tool-using AI agents: deterministic trace validation, policy-as-code trajectory rules, baseline-vs-candidate regression gates, and CI-ready JSON/Markdown/JUnit/SARIF reports.

Record & replay LLM API calls for deterministic agent tests.

Record, replay, and time-travel-debug LLM agent runs.

Offline, agent-oriented document exploration workspaces

Agent-RL Observability SDK — lightweight OpenTelemetry-based tracing

Lint AI agent instruction files such as AGENTS.md, CLAUDE.md, Cursor rules, and Copilot instructions.

ARC — Agent Runtime Cockpit: Python daemon, adapters, context retrieval

Action Commit Safety for AI agents: exact-call approval, UNKNOWN outcome recording, idempotent execution, and signed audit.

Never worry about your AI agent breaking production again. Kill the process mid-flight — the wire transfer is still safely refunded, your database stays clean, and your users never lose trust. agent-saga is the 1-line bulletproof safety OS that turns unpredictable LLMs into crash-proof, enterprise-grade autonomous engines.

OpenSandbox backend SDK for AI agents

Python client for agent-sanitizer: a thin bridge to the Node.js CLI for byte-identical sanitization verdicts.

Local-first cross-harness agent-session portability: export a live coding-agent session from one harness and resume it in another.

Local-first multi-agent coding session archive tooling — discover, export, and baseline your AI coding sessions

Multi-agent session reader (Claude, Codex, OpenCode, Antigravity, Pi) with MCP server

Offline integrity checks and non-destructive recovery bundles for Codex rollout JSONL

Optimize a Claude Code / Agent skill's description frontmatter for reliable triggering — no API key required.

Versioned evaluation, coverage, and release governance for Agent Skills

A Python framework for managing reusable AI skills throughout their lifecycle.