aped-debug

What problem does it solve?

It helps you troubleshoot failing software by forcing a disciplined loop: build a deterministic feedback signal, reproduce the exact failure, generate falsifiable hypotheses, instrument safely, then apply the smallest fix with a regression test and a clean post-mortem.

Core Features & Use Cases

• Disciplined diagnosis loop: Build a fast pass/fail feedback loop, reproduce the bug, hypothesize ranked causes, instrument one variable at a time, and iterate until the root cause is validated.
• 3-failed-fixes safeguard: Stops after three failed attempts to prevent endless patching when the underlying assumption is wrong.
• Evidence-driven cleanup & post-mortem: Ensures tagged debug probes are removed and commits capture cause + falsified hypotheses for future discovery.
• Invocation-context aware: Runs as standalone, from aped-dev (persistent test red), or from aped-review (root-cause finding routing).

Quick Start

Use the aped-debug skill when you see tests failing or you’re stuck debugging a production issue, and ask it to find the root cause from the exact failure description.