AI Agent Hub
Back to plugins
🧠

dsh-skill-trail

Memory Updated 2026.08.23

Run the following command in DeepSeek Harness:

dsh plugin install Bryan-cmf/dsh-skill-trail

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

Run dsh plugin install Bryan-cmf/dsh-skill-trail in the DeepSeek Harness terminal to install; source code is available at https://github.com/Bryan-cmf/dsh-skill-trail

About this plugin

In a DeepSeek Harness conversation it is often hard to tell at a glance whether a given skill was actually triggered or whether a tool call truly succeeded versus failed. dsh-skill-trail renders a compact trail beneath every assistant reply, surfacing what was triggered, what was used successfully, and where things went wrong in that turn.

The badges are not self-reported by the model. A host-side session-projection unit folds the committed session log with a pure function: each tool-call event becomes a triggered entry, and the paired result is matched by callId to settle a success or failure verdict. The trail is structurally verifiable, replayable, and strictly per-turn, so one reply never bleeds into the next and no extra isolation realm is required.

It suits anyone who wants to audit skill-invocation chains, verify tool execution outcomes, or demonstrate to colleagues exactly what a given turn did in a DSH workspace. Note that the author has since migrated further features to dsh-skill-ledger; this repository is kept as a design-history record.

Use Cases

  • Verify whether a given reply actually triggered a specified skill
  • Pinpoint the exact step where a tool call returned a failure
  • Show a team the full execution chain of a single conversation turn

Best For

  • Developers who need to audit skill-invocation chains
  • Engineers troubleshooting tool execution failures
  • Technical leads demonstrating DSH workspace execution details to teammates