Packages

A small async actor framework

Openagency public package coordinate scaffold

Small headless runtime that owns coding-agent processes and sessions

Async client for the AgencyProxy local and remote wire protocol

Versioned wire protocol shared by AgencyProxy clients and servers

Openagency public package coordinate scaffold

Agent Process Manager — orchestrate AI coding agents with PTY isolation, fleet management, and 32 MCP tools

Safe collections that can be mutated while being iterated through

A lightweight, stateful job queue for Rust backed by PostgreSQL with sqlx. Inspired by AgendaJS.

Terminal countdown clock for presentations, with full day agenda support

A flexible AI Agent SDK for building intelligent agents

Drive the Claude Code, Codex and GitHub Copilot CLIs headlessly from Rust. One request type, one event stream and one session model across all three, with resume and fork.

Data model and runtime adapters for authoring coding-agent assets (skills, agents, hooks, scripts, packs, MCP servers) once and exporting to Claude Code, Codex CLI, Gemini CLI, and other runtimes.

Ahp public package coordinate scaffold

Ahp public package coordinate scaffold

A Rust Framework to build TUI Agents

Core runtime for agent-air - LLM orchestration, tools, and permissions (no TUI dependencies)

TUI frontend for agent-air - ratatui-based terminal interface

agent-anacrusis

Ahp public package coordinate scaffold

Human-in-the-loop approval gate for LLM agent tool calls (Rust core + Rig adapter).

Rust support library for the Agent Approval Interchange Specification

Federated public Q&A protocol for AI agents — signed Q/A/Rating, content-addressed, pull federation (Rust port of @p-vbordei/agent-ask)

Deterministic knowledge base indexer for AI agents

agent-audience

A lightweight Agent Runtime Kernel for building AI agents in Rust

Lua-first Agent Runtime built on AgentMesh

Host runtime + Lua stdlib bridge + EventBus for agent-block

rmcp client wrapper for agent-block (MCP server lifecycle + custom handler)

In-process mock LLM servers and a provider wire-shape catalog (spec-conformant and deliberately broken tool-calling shapes) for testing LLM clients

Shared error types and observability utilities for agent-block

A powerful LLM-driven agentic EPUB book translator with robust progress monitoring and recovery

zerobox

Local-first CLI to read, compare, and hand off context across Codex, Claude, Gemini, and Cursor sessions.

Connect applications to installed AI agents

The capability leash for agent tools: a ready-to-use, capability-governed tool registry for the Gilamonster agent line.

Windows AppContainer process launcher for agent-bridle (agent-bridle#51, ADR 0009).

Layered configuration loader (defaults → file → env → API) for agent-bridle (agent-bridle#142, ADR 0017).

Capability-enforcement core for agent-bridle: the Tool trait, Gate (mint-token enforcement), Registry, and Caveats leash.

Close ambient file descriptors on confined spawn (agent-bridle#319). The one unsafe-permitting seam that keeps agent-bridle-core `forbid(unsafe_code)`.

Privileged Tier-1.5 mount-namespace jail for agent-bridle program-identity confinement (ADR 0013 D4 / #109).

MCP (Model Context Protocol) stdio server frontend over the agent-bridle capability-governed tool registry.

Capability-confined safe-subset, Brush, and host-shell engines for agent-bridle.

A capability-confined web_fetch tool for agent-bridle: the `net` enforcer (host allowlist + SSRF block + per-redirect re-check).

Fast browser automation CLI for AI agents

Unofficial Rust library crate for agent-browser core browser automation. Not affiliated with or maintained by Vercel.

Fork of agent-browser with rustls-tls-native-roots — works behind TLS-inspecting egress proxies (Anthropic Managed Agents, Cloudflare WARP, corporate MITM). Drop-in replacement; binary still named `agent-browser`.

Set threshold alarms on agent cost or token budgets

agent-cadence-progress

agent-call-response