Help us improve
Share bugs, ideas, or general feedback.
From plangate
Create a structured investigation document for complex bugs. Use when a bug requires more than a quick fix — tracks hypotheses, trials, root cause, and lessons learned.
npx claudepluginhub bishnubista/plangate --plugin plangateHow this skill is triggered — by the user, by Claude, or both
Slash command
/plangate:investigateThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
When a bug requires more than a quick fix, create a structured investigation document that tracks your debugging process. This prevents repeating failed approaches, documents root causes for future reference, and builds a knowledge base of project-specific issues.
Writes an engineering post-mortem for a fixed bug: root cause, mechanism, fix, validation, and how it slipped through. Refuses to draft if root cause or fix is unconfirmed.
Walks through evidence-first debugging: clarify expected behavior, reproduce, hypothesize, and plan fixes before changing code.
Orchestrates persistent, hypothesis-driven debugging sessions across conversations, delegating analysis to subagents and tracking in debug files.
Share bugs, ideas, or general feedback.
When a bug requires more than a quick fix, create a structured investigation document that tracks your debugging process. This prevents repeating failed approaches, documents root causes for future reference, and builds a knowledge base of project-specific issues.
/plangate:investigate {problem-name} — Creates or resumes an investigation doc
Create the file at:
docs/investigations/{YYYY-MM-DD}-{problem-name}.md
Ensure the directory exists:
mkdir -p docs/investigations
Use this template:
# Investigation: {Problem Name}
**Date:** {YYYY-MM-DD}
**Status:** Open
**Severity:** {Critical/High/Medium/Low}
## Problem Statement
{Clear description of the bug: what happens, what should happen, when it was first observed}
## Reproduction Steps
1. {Step 1}
2. {Step 2}
3. {Expected: ...}
4. {Actual: ...}
## Environment
- Stack: {from plangate-manifest}
- Branch: {current git branch}
- Last known working commit: {SHA if known}
## Hypothesis Log
| # | Hypothesis | Evidence For | Evidence Against | Status |
|---|-----------|-------------|-----------------|--------|
| 1 | {hypothesis} | {evidence} | {evidence} | Testing/Confirmed/Rejected |
## Trial Log
| # | Action | Expected | Actual | Result | Duration |
|---|--------|----------|--------|--------|----------|
| 1 | {what you tried} | {expected outcome} | {actual outcome} | ✅/❌ | {time} |
## Root Cause
{Filled in when root cause is identified}
## Fix Applied
{Description of the fix, with commit SHA}
## Lessons Learned
- {What would have helped find this faster?}
- {What pattern should be avoided?}
- {What should be added to testing?}
As you debug, update the document in real-time:
From Superpowers' systematic debugging:
When the fix is applied and verified:
Resolvedgit commit -m "docs: close investigation — {problem-name}"The investigation doc is not optional overhead. It prevents you from:
Don't batch updates. Record each hypothesis and trial as you go. If the investigation gets interrupted (context limit, session end), the document preserves all progress.
After closing an investigation, the Lessons Learned section is what prevents the same bug class from recurring. Be specific — "add integration tests for X" is better than "test more."