investigate

Trace runtime failures with file:line citations before proposing fixes.

1|Updated Oct 19, 2025
One-click install
npx skills add https://github.com/acartine/osterman --skill investigate-acartine
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: investigate
Source: https://github.com/acartine/osterman/tree/main/skills/investigate
Command: npx skills add https://github.com/acartine/osterman --skill investigate-acartine

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Evidence-gated debugging that requires citing file:line references and actual outputs before proposing fixes, reducing guesswork.

Core Features & Use Cases

  • Evidence collection: Reproduce the failure, collect runtime outputs, and trace code paths with explicit citations.
  • Evidence-based conclusions: Present root-cause hypotheses only when supported by cited evidence.
  • Approval-driven workflow: Gate potential fixes behind an explicit approval step to prevent premature changes.

Quick Start

Ask the agent to begin an evidence-first investigation and cite file:line references for every finding.

Frequently Asked Questions about investigate

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I find a root cause with evidence-based debugging instead of guessing?

Evidence-based debugging identifies a failure scenario and collects concrete evidence from runtime outputs, logs, and code paths. It traces execution with file:line references to substantiate findings, ensuring root-cause hypotheses are supported by cited evidence.

How do I trace an execution path with file:line references during an investigation?

Trace the execution path by reproducing the failure and collecting runtime outputs. Explicit file:line references are cited for every claim to substantiate findings, mapping the exact code paths traversed during the failure scenario.

When should I propose fixes during root-cause analysis?

Fixes should only be proposed after root causes are confirmed with cited evidence. The investigation phase strictly avoids code changes, gating potential fixes behind an explicit approval step to prevent premature modifications.

What is the best way to prevent premature code changes while debugging?

Use an approval-driven workflow that gates potential fixes behind an explicit approval step. This prevents premature code changes by requiring root causes to be fully confirmed with cited evidence like file:line references before any modifications occur.

Can I investigate runtime failures without reproducing the issue first?

No, reproducing the failure is a core requirement. Evidence collection begins by reproducing the failure scenario to gather actual runtime outputs and trace code paths, which are then used to substantiate root-cause hypotheses.

Why does my debugging workflow result in guesswork and unverified fixes?

Guesswork occurs when fixes are proposed without evidence. This workflow requires citing file:line references and actual outputs before proposing fixes, reducing guesswork by presenting conclusions only when supported by cited evidence.