Runs multiple upstream MCP servers behind one local process, keeping sessions persistent and exposing a compact, policy-aware surface for tools, resources, and prompts via HTTP, CLI, and MCP-native clients.
Exposes MCP tools that give desktop agents durable two-tier memory with recall and graph traversal, project code intelligence (symbol lookup, impact walking, notes search), and web search with page-to-Markdown conversion. It also runs a local model gateway that routes Anthropic- and OpenAI-shaped traffic across providers with usage accounting.
Fronts many downstream MCP servers through one session, presenting a unified tool namespace with collision resolution, fleet-wide allow/deny/hide policy, per-tool rate limits, centralized audit logging, Prometheus metrics, and bearer-key auth at the edge. Supports both stdio drop-in and Streamable HTTP deployment so agents point at a single endpoint instead of N servers.
Enables MCP clients to manage SQLite-backed TODOs and schedule durable coding-agent jobs over stdio, with persistent job memory, run history, and a read-only dashboard.
This MCP server gives an AI coding host controlled, client-agnostic access to a local machine's developer capabilities: transactional file edits, shell and PTY execution, Git and project inspection, language-server operations, isolated browser automation, Windows UI automation, VS Code state, durable jobs, and checkpoints, all bounded by filesystem root allowlisting and resource admission controls. It speaks standard MCP over stdio, Streamable HTTP, and legacy HTTP+SSE, and can also run headless as a long-lived job runner without any hosted model vendor.
Provides a centralized, deterministic MCP endpoint with routing, logging, and browser automation, enabling multiple LLMs to share a consistent set of tools including jq, ripgrep, Git, Python, and headless Chromium.