debug-investigation

Investigate bugs and incidents through evidence gathering, hypothesis testing, and root cause analysis.

Updated Jul 8, 2026
One-click install
npx skills add https://github.com/Zhachory1/autopraxis --skill debug-investigation-zhachory1
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: debug-investigation
Source: https://github.com/Zhachory1/autopraxis/tree/main/skills/debug-investigation
Command: npx skills add https://github.com/Zhachory1/autopraxis --skill debug-investigation-zhachory1

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Debugging production issues, failing tests, and unclear bugs often devolves into log fishing, untested guesses, and endless reproduction attempts. This Skill enforces a disciplined investigation loop: define the symptom precisely, gather evidence before theorizing, test hypotheses against a ledger, and stop on a confirmed root cause or escalate with a ruled-out list. ## Core Features & Use Cases - Structured Symptom Definition: Captures expected-vs-actual behavior, blast radius, frequency, and done criteria before any investigation begins. - Bounded Hypothesis Loop: Tests candidate root causes against logs, traces, metrics, and code paths, maintaining a confirmed/ruled-out/inconclusive ledger with budget caps to prevent infinite loops. - RCA and Handoff Packaging: Produces a root cause analysis with fix, prevention actions, and escalation paths to councils or human approval gates for high-risk or ambiguous cases. - Use Case: A developer facing a failing integration test in CI runs the workflow in lite mode to define expected-vs-actual behavior, trace the implicated code path, and confirm the root cause with a targeted fix, all within one hypothesis loop. ## Quick Start Investigate this failing test by defining the symptom, gathering evidence from logs and recent changes, and confirming the root cause before proposing a fix.

Frequently Asked Questions about debug-investigation

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

FAQPage Schema
How do I debug a production incident systematically?▼

Start by defining the symptom precisely: expected versus actual behavior, blast radius, and frequency. Then gather evidence from logs, traces, metrics, and recent deploys before forming hypotheses, and test each candidate root cause against the evidence until one is confirmed.

How to find the root cause of a failing test?▼

Use a bounded hypothesis loop: capture expected versus actual output, trace the implicated code path, and confirm or rule out candidate causes with targeted checks. Keep a ledger of ruled-out hypotheses so the investigation never repeats dead ends.

What should a root cause analysis document include?▼

An RCA should record the confirmed root cause with supporting evidence, contributing factors, the applied fix, prevention actions such as tests and alerts, and any open risks. It should also list hypotheses that were ruled out during the investigation.

What do I do when a bug cannot be reproduced locally?▼

Set a reproduction budget cap and rely on production evidence such as logs, traces, and metrics instead of looping forever. If the budget is exhausted, stop and report the evidence gathered plus the ruled-out causes, then escalate to a human owner.

When should a debugging investigation escalate to human review?▼

Escalate when there is production blast radius, evidence exhaustion, a high-risk fix, conflicting hypotheses, or an ambiguous root cause. Security, privacy, or reliability concerns also justify escalation to a council or human approval gate.