AI Agent Hub
Back to plugins
⚙️

ci-runner

Workflow Updated 2026.08.25

Run the following command in DeepSeek Harness:

dsh plugin install JohnXu22786/ci-runner

Paste the following prompt into your AI chat to install this plugin:

Run dsh plugin install JohnXu22786/ci-runner inside your DeepSeek Harness instance to install the plugin; source code is available at https://github.com/JohnXu22786/ci-runner.

About this plugin

Running a CI pipeline or a local test suite is only half the battle — the real time sink is wading through thousands of lines of log output after a failure to pin down the cause. ci-runner collapses that loop into a single call: it dispatches a GitHub Actions workflow or a local pipeline (npm test, pytest, etc.), polls until the run finishes, hands the log tail to DeepSeek, and returns a structured Markdown diagnostic report covering failure stage, error classification, most likely root cause, suggested fix steps, and related files.

Security and control are baked into the design. The GitHub token is read from an environment variable at call time and never persisted or logged. Every returned log is masked against known secrets. Local commands are hard-killed on timeout and capped in per-stream output so a runaway process cannot blow up the agent context. The plugin ships with zero runtime dependencies, is plain Node.js ESM, and exposes five tools (ci_trigger, ci_status, ci_logs, ci_watch, ci_diagnose) plus a matching CLI, both sharing a single CiService instance for identical behavior.

Built for developers working inside a DeepSeek Harness agent workflow who want to close the CI feedback loop automatically — so the agent does not just trigger and wait, but reads the failure, understands it, and suggests the fix.

Use Cases

  • Dispatch a GitHub Actions workflow, poll to completion, and stream masked logs back
  • Run a local pipeline like npm test or pytest and auto-diagnose failures via DeepSeek
  • Close the run-read-diagnose loop in a single agent workflow call

Best For

  • Developers building automated agent workflows in DeepSeek Harness
  • Engineers who want to cut post-failure debugging time to near zero
  • DevOps teams wiring local test pipelines into agent-driven diagnosis