Project-family orchestration workflow: same migration, ralph single-repo loop, multi-project dispatch; optional session team engines (coordinate / lifecycle / swarm).

A skill-driven specification framework and ACP cockpit

MCP server for xAI Grok APIs - image generation, chat, vision, search, and video

Autonomous coding agent CLI. Supports 8 providers: Groq, Gemini, Claude, DeepSeek, OpenAI, Mistral, xAI Grok, OpenRouter. Understands codebases, writes files, runs commands, searches the web.

ACP agent bridge for intentic — drive the agents in your intentic sandbox (Claude, Codex, Grok, any ACP agent) from Zed, JetBrains, or any ACP editor

Token-saving proxy for Claude Code: renders bulky context (system prompt, tool docs, old history) as dense PNGs to cut input tokens. Runs on Node and Cloudflare Workers.

Relay roles for Pi: run any Pi subagent on an external coding agent (headless Claude Opus via `claude -p`) through a provider seam, driven by the subagent's `model` field.

Personal AI agent conventions - installs shared guidelines into a repo's CLAUDE.md (Claude Code) and AGENTS.md (Codex, Cursor) from a single source of truth.

MCP server for Bandi-SpyWebSocket — Paper console, players, permissions via WebSocket

Compact CLIProxyAPI account usage meters for Pi Coding Agent

X search for any Pi model through xAI or your own Responses API gateway, with zero runtime dependencies.

Visual creative expert plugin — search inspiration, enhance prompts, and generate AI images with intelligent workflow orchestration

MCP server for routing image generation and editing across xAI, Google Gemini/Nano Banana, and OpenAI.

Use Grok Build from Pi as a managed collaborator for review, research, edits, and media.

Persistent tmux terminal MCP for launching and driving Claude, Codex, Grok, or Composer from any MCP client, cross-vendor or same-vendor. Also runs durable PTY sessions for SSH, containers, and REPLs.

1agents ACP WebSocket bridge-server (wraps acpx runtime for Chat UI)

Grok plugin for capability-gated, read-only second-opinion reviews through supported external coding harnesses, with a host scope gate and bounded subprocess execution.

Browse and expose local Codex, Claude Code, Grok, Kiro, Qoder, Cursor, Antigravity, MiniMax Code, and OpenCode conversation history through a local UI and stable data API.

Deploy Forward tracker — ranks agentic engineers on what they ship. Measures your AI build time, tokens and outcomes locally (Claude Code; Codex next). Metadata only; never your code or prompts.

Stateful Node.js sessions for Codex, Claude, Grok, Bedrock, and Responses-compatible model providers.