debug

Systematic Debug

Ultrathink!

Systematically investigate and diagnose the issue using evidence-based debugging.

Parse arguments from: $ARGUMENTS

• If arguments contain --deep, enable deep mode (agent-assisted debugging) and strip --deep from issue description
• Remaining text is the issue description (error message, bug description, or symptom)
• If no arguments provided: analyze the current conversation context to identify the error, bug, or unexpected behavior

---

Step 1: Evidence Gathering

Gather targeted evidence about the issue. This step runs in both default and deep modes.

1a. Parse the Issue

Extract structured information from the issue description or conversation context:

• File paths mentioned in error messages or stack traces
• Line numbers from stack traces
• Error types/names (e.g., TypeError, NullPointerException, panic)
• Function/method names referenced in the error

1b. Git Context

Run these commands to understand the current state:

!git status --short !git diff --stat !git log -5 --oneline

If specific files were identified in the error, run git diff -- <file> and git log -3 --oneline -- <file> for each relevant file.

1c. Ecosystem Detection

!ls package.json pyproject.toml build.gradle build.gradle.kts pom.xml Cargo.toml go.mod Makefile 2>/dev/null || true

1d. Error Reproduction

Detect the test runner from the ecosystem and attempt to reproduce the error:

Node.js (if package.json exists): !node -e "try{const s=require('./package.json').scripts||{};console.log(JSON.stringify({test:s.test||null}))}catch(e){}" 2>/dev/null

• If a specific test file is mentioned in the error: run only that test
• If a source file is mentioned: find and run its corresponding test file
• Otherwise: run the full test suite

Python (if pyproject.toml exists): !grep -c '\[tool\.pytest' pyproject.toml 2>/dev/null && echo "pytest detected" || true

• Run pytest with the specific test file or function if identifiable