Packages

Rust SDK and CLI for the AgentBlogs API

x402 API gateway for Agentbot agents to access Tempo network payments, colony membership, fitness scoring, and dynamic pricing via borg-0's x402-node

zerobox

Command-line client for agentbus, an MCP-native message bus for AI agents.

Core types for agentbus: envelopes and the spool store (registry, delivery, asks, event log).

stdio MCP bridge for agentbus, exposing the message bus as MCP tools.

agentbus daemon: HTTP/REST and SSE message bus server for MCP-capable AI agents.

Unit tests for AI agents. Run behavioral tests in CI, score with an LLM judge, and export signed evidence your auditors accept.

Structured-output enforcer for LLM responses. Repair + validate + (optional) retry-with-LLM. BYO-LLM, BYO-schema.

Universal stream formatter for AI coding agent output

A CLI tool for browser automation via the Chrome DevTools Protocol

A standard way to declare and run agent containers for your projects.

Rust implementation of the AgentContract specification — behavioral contracts for AI agents

A minimal Rust crate that gives any application agentic capabilities.

Tiny control-plane primitives for AI agents.

Unified auth pool and LLM API client for Claude Max Plan, OpenAI, and more

agentctx

Agent daemon for secure capability execution with pluggable isolation backends

Interface for agent persistence.

Reserved crate name for the future agentdev project.

Audit and trace autonomous AI code contributions in git repositories

Virtual file tree infrastructure for agent-ready file layouts using CoW reflinks

CLI for agentdir — virtual file tree infrastructure for agent-ready file layouts

Project-scoped AI agent and plugin environment manager

Core library for agentenv - project-scoped AI agent and plugin environment manager

Generate production-ready MCP (Model Context Protocol) servers and clients from OpenAPI specs

Agenterra fork of Rust SDK for Model Context Protocol

Agenterra fork of Rust SDK for Model Context Protocol macros library

Unified agent file installer for Claude Code, OpenCode, Codex, and Cursor

Fit messages to an LLM context window. Token-aware truncation with pluggable tokenizers and multiple strategies.

Git worktree workflow manager with AI agent task monitoring, Kanban board, and terminal UI

Shared types, configuration, themes, and error handling for the Flow workspace manager

A Rust crate for building multi-agentic applications.

REST API server for AgentForge (F-07)

Benchmark comparison: runs agents against GAIA, AgentBench, and WebArena tasks and reports percentile vs. published baselines (v2 F-05)

CLI for AgentForge (F-08)

Shared types, data models, and error types for AgentForge

Database layer: SQLx queries and migrations for AgentForge

Fine-tuning dataset exporter: converts labeled traces to OpenAI, Anthropic, and HuggingFace JSONL formats (v2 F-02)

Promotion gatekeeper: score gate, regression gate, stability gate (F-06)

Multi-agent graph testing: runs agent teams (CrewAI-style, LangGraph-style) as composed units with per-node scoring (v2 F-03)

Observability hooks: exports AgentForge traces to OTLP, LangSmith, and Datadog backends (v2 F-06)

Online shadow evaluation: routes live traffic to champion and candidate simultaneously and compares results (v2 F-01)

Optimizer: mutation strategies to generate candidate agent variants (F-05)

Agent file parser and schema validator (F-01)

Red-teaming mode: adversarial safety probing with jailbreak, prompt injection, data leakage, and constraint bypass scenarios (v2 F-04)

Agent execution engine: parallel workers, trace capture, retry logic (F-03)

Scenario generation (F-02): schema-derived, adversarial, domain-seeded

Trace analyzer and scorer (F-04): deterministic assertions, LLM-as-judge, failure clustering

Runtime protection and security monitoring for AI agents — The CrowdStrike for AI Agents