BailingHub MCP Server
Enables submission, tracking, and retrieval of governed tasks from DeepSeek Harness against BailingHub routes, allowing job management without opening embedded chat interfaces.
README
BailingHub for DeepSeek Harness
简体中文 | English
Run governed tasks from DeepSeek Harness against a business route already connected to BailingHub, then track the same job and read its available result. You can do this from the Harness Web UI on your own computer without opening the embedded chat page in each business backend.
This is an independent community integration. It is not developed, certified, endorsed, or recommended by DeepSeek.
What You Get
After installation, DeepSeek Harness discovers three native tools:
| Harness tool | What it does |
|---|---|
mcp__bailinghub__submit_governed_job |
Submit task text with a stable request ID |
mcp__bailinghub__get_governed_job |
Read the current state and public result of one job |
mcp__bailinghub__wait_for_governed_job |
Wait briefly for the same job without resubmitting it |
BailingHub remains the control plane between the model and the business system. The URL, Client Token, and route are operator configuration, not model arguments. The model cannot switch to another route or supply administrator, executor, approval, or business-system credentials through these tools.
Install
Prerequisites:
- Node.js
22.19.0+or24+; pnpmand@deepseek-ai/dsh@0.1.0-rc.7;- a reachable BailingHub deployment;
- one BailingHub Client Token restricted to the route you want this Harness profile to use.
npm install --global pnpm @deepseek-ai/dsh@0.1.0-rc.7
dsh plugin --profile web add dsh-bailinghub@0.1.1
Configure the process that starts Harness:
export BAILINGHUB_BASE_URL='https://hub.example.com'
export BAILINGHUB_CLIENT_TOKEN='replace-with-a-route-scoped-client-token'
export BAILINGHUB_ROUTE='order_assistant'
Then verify the composed profile and start the local Web UI:
dsh --profile web --dump-config
dsh web
The default Web UI is local to your computer. Ask Harness to submit a task through
BailingHub, preserve the returned job_id, and wait for or query that same job.
For a first try, use a task supported by your configured route and say explicitly: submit it
once with a stable request_id, preserve the returned job_id, and query that same job if a
bounded wait times out.
Correct Task Flow
- Generate one stable
request_idfor one business request. - Call
mcp__bailinghub__submit_governed_jobonce. - Preserve its
job_id. - Call
wait_for_governed_job, or callget_governed_joblater. - A wait timeout is not a failed task. Do not submit a replacement request.
Identity and Permission Boundary
Using the local Harness UI removes the need to open a business backend's embedded chat UI;
it does not turn the Harness login or local user into a trusted business identity.
The current 0.1.x bundle deliberately does not accept an acting subject as tool input. A route or
downstream business system must still resolve trusted identity and perform final
authorization. If the selected action requires identity that the configured path cannot
establish, it should remain unavailable or be rejected.
This bundle governs only tasks submitted through its three BailingHub tools. It does not intercept or govern every other tool installed in DeepSeek Harness.
Supply-Chain Note
This bundle contains no custom runtime JavaScript, no production dependencies, and no
install scripts. When Harness starts the bundle, its built-in MCP Client runs the pinned
command npx -y --package=bailinghub-mcp-server@0.1.1 bailinghub-mcp-server outside the
agent sandbox. The first start may
need npm network access. Review and pin the package before using it in a sensitive
environment; production images may pre-cache the exact version.
Non-loopback HTTP is rejected by the MCP server by default. Only set
BAILINGHUB_ALLOW_INSECURE_HTTP=true on a controlled private network where TLS terminates at
another trusted boundary.
Compatibility and Feedback
The first release is verified against DeepSeek Harness 0.1.0-rc.7,
bailinghub-mcp-server@0.1.1, and BailingHub Client API v1. Harness is still a developer
preview, so each Harness release requires a compatibility smoke test.
Report problems through GitHub Issues. Never include tokens, private deployment URLs, personal information, or production business payloads.
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.