Reproduce first: Run the failing case and capture the exact error. Don't guess — observe.
Read the stack trace bottom-up: The root cause is usually near the bottom. Work upward to understand the call chain.
Form a hypothesis, then test it: Don't shotgun-debug. Identify one likely cause, verify it, then move on if wrong.
Check the boundaries: Most bugs live at integration points — API calls, file I/O, type conversions, serialization. Check inputs and outputs at each boundary.
Verify the fix: Run the original failing case again. Then check that you didn't break the happy path.

Approach:

Start with grep or glob to find the relevant code
Read the error logs or stack trace carefully
Add targeted debug output if needed (remove it after)
Fix the root cause, not the symptom
Run tests to confirm

Más de este repositorio

mismo repositorio

agent-plane-knowledge

dbczumar/agent-plane

Deep reference on agent plane config format, executor types, skill/tool structure, and conventions. Load when you need to look up how the platform works.

2026-04-193

generate-agent

dbczumar/agent-plane

Patterns and templates for generating valid agent plane agent directories. Load when ready to create files.

2026-04-193

repl-sdk

dbczumar/agent-plane

Build terminal REPLs for agent-plane using the Python UI SDK. Use when the user asks about building a REPL, terminal chat, or CLI frontend for an agent.

2026-04-193

deploy-databricks-app

dbczumar/agent-plane

Deploy agent-plane to Databricks Apps with Lakebase and UC Volumes. Use when setting up infrastructure, deploying, redeploying, troubleshooting, or connecting to a remote agent-plane instance on Databricks.

2026-04-153

detect-framework

dbczumar/agent-plane

Detect Python agent frameworks from code imports and map them to agent plane executor types. Load when the user has existing agent code to integrate.

2026-04-123

deep-research

dbczumar/agent-plane

Investigate a topic in depth, find sources, and synthesize findings.

2026-04-103

name	systematic-debugging
description	Systematically diagnose and fix bugs using logs, stack traces, and targeted investigation.

When debugging:

Reproduce first: Run the failing case and capture the exact error. Don't guess — observe.
Read the stack trace bottom-up: The root cause is usually near the bottom. Work upward to understand the call chain.
Form a hypothesis, then test it: Don't shotgun-debug. Identify one likely cause, verify it, then move on if wrong.
Check the boundaries: Most bugs live at integration points — API calls, file I/O, type conversions, serialization. Check inputs and outputs at each boundary.
Verify the fix: Run the original failing case again. Then check that you didn't break the happy path.

Approach:

Start with grep or glob to find the relevant code
Read the error logs or stack trace carefully
Add targeted debug output if needed (remove it after)
Fix the root cause, not the symptom
Run tests to confirm