Skip to content

jiezeng2004-design/dsh-chatgpt-bridge

50Last commit Aug 15, 2026

dsh-chatgpt-bridge DSH plugin

dsh-chatgpt-bridge is a standalone DSH plugin that exposes 15 MCP tools over Streamable HTTP or stdio. It lets ChatGPT Web create, view, continue, and supervise DSH agent sessions without modifying the DSH core. The bridge uses DSH's public plugin seams (AgentRegistry, SessionStore, approval waterfall, etc.) and runs in the same web profile for live session streaming.

How to install the dsh-chatgpt-bridge DSH plugin

dsh plugin --profile web add dsh-chatgpt-bridge

Copying does not run this command. Review the repository and version before installing the dsh-chatgpt-bridge DSH plugin.

dsh-chatgpt-bridge DSH plugin data source

dsh-chatgpt-bridge DSH plugin snapshot date: Aug 16, 2026

discovered

What the dsh-chatgpt-bridge DSH plugin can do

  • Exposes 15 MCP tools: dsh_health, dsh_list_workspaces, dsh_start_goal, dsh_send_message, dsh_approve, dsh_update_goal, dsh_wait_goal, dsh_create_session, and more.
  • Supports Streamable HTTP (default http://127.0.0.1:3456/mcp) and stdio transports.
  • Authenticates via Bearer token (generated or user-set via DSH_CHATGPT_BRIDGE_TOKEN).
  • Allows ChatGPT to create, resume, list, and cancel DSH sessions, and to send messages and approve decisions.
  • Integrates with DSH's approval waterfall and user question provider for interactive supervision.

Where the dsh-chatgpt-bridge DSH plugin fits

  • Use ChatGPT to perform read-only health checks and workspace listing on a local DSH instance.
  • Create a supervised goal (e.g., 'read-only inspect project') and monitor its progress through ChatGPT.
  • Let ChatGPT create and manage DSH agent sessions that modify files inside registered workspaces under DSH's sandbox policy.
  • Automate DSH workflow approvals or answer user questions directly from the ChatGPT interface.
  • Resume persisted DSH sessions across ChatGPT conversations using the bridge's session persistence integration.

Who the dsh-chatgpt-bridge DSH plugin is for

  • DSH users who want to control agent sessions from ChatGPT Web.
  • Developers building AI-assisted workflows that combine ChatGPT's natural language with DSH's agent execution.

dsh-chatgpt-bridge DSH plugin limitations

  • Requires a working DSH installation (dsh on PATH) and Node.js >= 22.
  • ChatGPT Web cannot directly connect to localhost MCP endpoints; a secure tunnel (e.g., OpenAI's supported Secure MCP Tunnel) is needed.
  • The bridge only binds to 127.0.0.1 (loopback) and never exposes a public interface by itself.
  • Session live streaming in DSH Web only works if the bridge and Web UI run in the same profile/runtime.
  • Workspaces must be pre-registered in DSH; the bridge never auto-registers arbitrary paths.

dsh-chatgpt-bridge DSH plugin: from the repository README

Quoted from the jiezeng2004-design/dsh-chatgpt-bridge README, the upstream source of the dsh-chatgpt-bridge DSH plugin. Copyright remains with the original authors.

An MCP bridge that lets **ChatGPT Web** create, view, continue and supervise **DeepSeek Harness (DSH)** agent sessions through the official **Model Context Protocol**. v0.3.0 — *Goal Control Plane*. The bridge only *connects* — DSH keeps its own session log, agent loop, tools, skills, subagents, workflows, approvals, sandbox and workspace security model. It is a standalone DSH plugin: **zero DSH core modifications**. > Self-hosted / dogfooding development: implemented against the installed DeepSeek > Harness source (`0.1.0-rc.6`) and verified end-to-end against a live local DSH > runtime with the official MCP SDK client. --- # Quick Start This gets a new user from zero to a verified ChatGPT ↔ DSH connection. Deep architecture and configuration details follow below — you do not need them to install and verify. ## Requirements - **Node.js >= 22** installed and on your `PATH`. - **A working DeepSeek Harness (DSH) installation** — `dsh` on your `PATH` (or use `pnpm dlx @deepseek-ai/dsh@0.1.0-rc.6` in place of `dsh` in every command below). - **A web profile is recommended.** The Web UI and the Bridge should run in the same web profile/runtime so that ChatGPT-created session

Read the full READMERepository license: MIT

dsh-chatgpt-bridge DSH plugin questions

ChatGPT cannot connect to the bridge. What should I do?

The bridge listens on http://127.0.0.1:3456/mcp, which is a loopback address. ChatGPT Web cannot reach it directly. You need to use a secure tunnel (e.g., OpenAI's currently supported Secure MCP Tunnel) that forwards requests to the local endpoint. Make sure the tunnel sends the correct Bearer token from step 3 of the Quick Start.

I get a 401 Unauthorized error when connecting.

Read the token from the file `$HOME/.dsh/chatgpt-bridge.token` (or the path shown in the README). The connector must send it as the `Authorization: Bearer <token>` header. Make sure the token matches the currently running bridge instance. If you set the `DSH_CHATGPT_BRIDGE_TOKEN` environment variable, the bridge uses that instead.

dsh_health works but no workspace appears in dsh_list_workspaces.

The bridge only lists workspaces that are already registered in DSH. It never auto-registers arbitrary paths. You must register the workspace in DSH (e.g., through the Web UI workspace settings or DSH configuration) before the bridge can see it.

Sessions created via ChatGPT do not appear live in DSH Web.

The bridge and DSH Web must run in the same web profile/runtime. If you run them as separate processes, sessions persist but will not stream in real time. Use a single `dsh web` command that loads both the Web UI and the bridge plugin.

The tool count is stale or dsh_update_goal is missing after upgrading.

After upgrading the plugin and restarting the profile, you need to re-scan or refresh the MCP tools on the ChatGPT side. v0.3.0 exposes 15 tools; `dsh_update_goal` is the 15th. If it's missing, force a refresh in the ChatGPT connector settings.