"Pix" matching MCP servers:

An MCP bridge for the Pi coding agent running in WSL, enabling task execution, research, review, and session management through MCP.

Enables web search for the pi coding agent using z.ai GLM native search with a DuckDuckGo fallback.

Enables starting, monitoring, stopping, and resuming detached agent runs with per-pool concurrency limits and run history.

Enables an assistant to run pi coding-agent tasks in the background: dispatching a task, polling its status and tool trace, reading the final result, and cancelling it when no longer needed, all without pulling the delegate's context into its own conversation.

Enables Codex to supervise one or more local Pi coding-agent workers through bounded task execution, state polling, steering, aborting, and listing. Tasks can run in a direct working directory or isolated Git worktree while Codex remains the planner and reviewer.

Manages MCP server configurations and exposes their tools to the Pi agent through a single gateway, with on-demand startup and discovery.

Enables Codex to delegate subtasks to concurrent background coding agents that can each run on any model provider supported by Pi, such as DeepSeek, Anthropic, Gemini, OpenRouter, or local models. Exposes MCP tools for listing models, spawning and waiting on agents, sending follow-ups, and aborting, backed by a live local web panel showing status, token usage, and streaming output.

Enables the Pi coding agent to drive Chrome through DevTools, providing browser automation tools, persistent or isolated profiles, screenshots, page analysis, and bundled agent policies.

Enables a coding agent such as Codex or Claude Code to delegate work to a local Pi coding-agent session, including second-opinion reviews, bulk reading kept out of the main context, and follow-up turns that retain context. Operators control what is possible by defining workspace and execution-profile aliases in a local config file, so clients cannot choose directories, models, providers, or permission levels themselves.

Enables Pi models to discover and activate MCP tools by search, making every MCP tool callable as a native Pi tool.

Enables MCP-aware agents to drive Modal serverless compute, including GPU workers and headless Chromium for web browsing with screenshots.

Composes arbitrary MCP tools using JavaScript, enabling discovery, composition, and execution of other MCP servers through a single 'exec' tool.

Enables a main agent to delegate tasks to reusable Pi agents with persistent sessions, question handoffs, bounded waiting, and in-run correction, while continuing other work. It supports named agents, context retention across MCP restarts, and result retrieval through MCP without repeated polling.

Exposes pi as a sub-agent for MCP-capable coding assistants, enabling task dispatch, messaging, and interactive dialog handling via tools like pi_dispatch, pi_send, and pi_respond.

Enables Codex to orchestrate persistent Pi RPC workers over a local MCP server, spawning dev, explore and review agents in isolated git worktrees or in place. Codex can inspect status, activity, token/cost metrics and real diffs, then merge only the accepted work and clean up idle workers.

Enables running isolated main and sub agent tasks in networkless Docker capsules with explicit input grants, verified artifact delivery, and independent reviewer oversight via MCP tools.

Enables Claude Code to query Pi through a JSON protocol with nesting, hop limits, token budgets, session lanes, and cancellation, treating peer output as untrusted tool results.

Exposes Pi execution nodes to any MCP client so it can dispatch sub-tasks to local or cloud models, poll and cancel them, and get command-based acceptance, diff-based evidence chains and automatic escalation to stronger models on failure.

Orchestrates multiple coding agents at once so an MCP client can delegate implementation work while retaining judgment: each agent runs in its own isolated git worktree and branch, claims the files it edits, and communicates via a mailbox and shared board. Supports blocking questions to the orchestrator, Codex-written acceptance tests locked before workers start, scope validation, and reporting on first-pass success.

Enables Claude Desktop and Claude Code to connect to multiple WLAN Pi devices through a single local stdio MCP interface, aggregating each device's tools, resources, and prompts with certificate pinning and token handling.