Local Codex Bridge
An MCP control plane for Codex native sessions, translating eight explicit operations with strict security boundaries for authenticated access through OpenAI Secure MCP Tunnel.
README
Local Codex Bridge
Frozen experimental reference. This repository is intended to remain read-only and is not production-ready or supported. It did not complete authenticated Tunnel, macOS real-device, or Android ChatGPT Work validation. See EXPERIMENTAL.md.
Local Codex Bridge is a narrow MCP control plane for Codex native sessions on a user-approved Windows or macOS host. It is designed for authenticated access through OpenAI Secure MCP Tunnel.
The bridge translates eight explicit operations. Codex native threads remain the only source of truth for session identity, history, runtime state, and results.
Status
Development is frozen. The source is published only to preserve the protocol, security-boundary, and testing work as an experimental reference. No runnable release artifact is provided.
Security boundary
- No arbitrary shell, process, filesystem, archive, delete, fork, or rollback tool is exposed.
- Callers choose a locally approved workspace identifier, never an arbitrary machine path.
- Callers cannot override the configured permission profile, sandbox, network policy, model-provider configuration, or App Server configuration.
- The bridge has no session database, transcript mirror, task registry, or durable pending-request store.
- Secrets belong in Windows DPAPI-protected storage or macOS Keychain and must not enter repository files, command arguments, logs, screenshots, or fixtures.
See the threat model and protocol decisions before operating the bridge. The architecture notes describe state ownership, the fixed process boundary, and restart behavior. Platform setup documents credentials, launchers, managed tunnel supervision, and recovery. The real-device acceptance runbook defines the Windows, macOS, Tunnel, and Android evidence procedure without retaining native identities or private host metadata beyond its explicit public-version allowlist. Its sanitized report format preserves a historical, uncompleted runtime-release gate; it is not evidence for this frozen source publication. The release runbook defines the fail-closed publication gate.
Development
Requires Node.js 24 or newer, npm, Codex CLI, and gitleaks.
npm ci
npm run check
npm run secret:scan
The commands above are for source inspection and test reproduction, not an installation or deployment recommendation.
Non-goals
V1 does not wake dormant ChatGPT conversations, implement Workspace Agents API triggers, control turns owned by another App Server runtime, or provide a general remote shell.
License
MIT.
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.