dsh-streaming-mcp-bridge
Run the following command in DeepSeek Harness:
dsh plugin install yabolee-kkk/dsh-streaming-mcp-bridge
Paste the following prompt into your AI chat to install this plugin:
Run dsh plugin install yabolee-kkk/dsh-streaming-mcp-bridge in DeepSeek Harness to install this plugin, sourced from https://github.com/yabolee-kkk/dsh-streaming-mcp-bridge
About this plugin
In Harness' ACP channel, external clients typically receive only the final answer while the chain of thought, tool invocations, and interim status transitions remain invisible. dsh-streaming-mcp-bridge closes that gap by bridging Harness' internal session.event and session.status streams outward, delivering thinking steps, tool calls, and state changes incrementally as they happen.
The plugin offers two operational modes. In MCP mode it uses Streamable HTTP transport and continuously emits notifications/progress while dsh_run_stream is executing, letting any MCP client follow the full task as a live event stream. In ACP mode it targets cc-connect and Feishu workflows, pushing each thought, tool input and result, and status update into Feishu messages in real time, with every item rendered as its own block and the final answer placed outside the code formatting for clean readability.
It is well suited for MCP client developers who need real-time observability, teams that want to track agent execution through Feishu, and integrations built on ACP that call for process transparency. By default it binds to 127.0.0.1 and requires Bearer token authentication, striking a balance between open capability and security.
Use Cases
- MCP client developers subscribe to agent thinking, tool calls, and status events to follow a task step by step
- Teams track every agent reasoning step and tool result in Feishu via cc-connect in real time
- ACP integrations that need intermediate state transitions, not just the final answer, to drive downstream logic
Best For
- Developers building streaming MCP clients on the MCP protocol
- Enterprise app teams wiring agent execution into Feishu or Lark
- Platform engineers who need process-level observability for Harness agents
Related Plugins
A method pack that makes AI coding agents plan against your real baseline, prove completion with fresh evidence, and reduce reworks and unsafe changes.
Turns the DeepSeek Harness session into a captain that builds a durable sub-agent team, splits goals into dependency-aware tasks, and coordinates work via direct messages and a live Web UI.
Gives coding agents design judgment, letting Claude Code, Cursor, and 70+ agents generate and iterate high-quality UI, presentations, and graphics right from the terminal.
Run the Pi ecosystem's plugins on DeepSeek Harness, unmodified, via a compatibility layer that implements Pi's public extension ABI on DSH's native services.