Skip to content
Development
Skill

/gather-evidence

Use when handed one factual question about the code or a running system. Returns evidence. Never proposes a fix.

From plugin
pane
48842 skills10 agents35 commands
Install
$ npx -y skills add dcouple/Pane --skill gather-evidence --agent claude-code

How it fires

How this skill gets triggered: by you, by Claude, or both.

  • Fires itselfAuto-invocation. Claude auto-loads it when your prompt matches the work.Auto-invocation is when the right skill fires by itself at the right moment, driven by a FLOW.md router and a hook, instead of you invoking it by name. It is the difference between a skill being installed and a skill actually getting used.Read the full definition →
  • You can call itInvoke it directly when you want it.
  • Slash command/gather-evidence

Context preview

The summary Claude sees to decide when to auto-load this skill.

Use when handed one factual question about the code or a running system. Returns evidence. Never proposes a fix.

SKILL.md

gather-evidence.SKILL.md
name: gather-evidence
description: Use when handed one factual question about the code or a running system. Returns evidence. Never proposes a fix.

Gather evidence

You answer one question with facts. The agent that asked will do the interpreting.

Method

1. Restate the question in one line. If it is really several questions, answer the first and list the rest. 2. Find the likely files with search and directory listings. Start from entry points: routes, handlers, exports, hooks, public interfaces. 3. Trace data flow and control flow only as far as the question needs. 4. For a failure: try to reproduce it. Find the failing test or write down the exact steps. Collect the error, the log lines, and the request or record involved. Compare a broken path with a working one when you can. Check recent history of the files involved. 5. Read files before making claims. Quote evidence directly.

Rules

  • Leave the code as it is. If diagnostic logging would settle it, say exactly what you would add and why, and stop.
  • Back every claim with a file reference, a command and its output, or a log line.
  • Report facts only. Improvements, fixes, and designs, even obvious ones, are for the agent that asked.
  • Stop as soon as the question is answered.

Report

  • The answer, in two or three sentences
  • Evidence: file and line references, commands with output, log excerpts
  • What you could not determine, and what would settle it
  • Confidence: high, medium, or low
Read more
Ships withpane

pnpm dlx runpane@latest

Get the whole plugin

Other skills on pane.