Menu
Guides systematic root-cause debugging. Use when tests fail, builds break, behavior does not match expectations, or any unexpected error appears.
Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.
Based on SOC occupation classification
Build a personalized learning roadmap with milestones and practice checkpoints
Structured document review for structure, clarity, and technical accuracy. Use for "review this doc", feedback, critique, or /doc-review.
Render a documentation-style Cursor Canvas that organizes architecture notes, API references, walkthroughs, and how-tos into a navigable layout with sections, tables of contents, and cross-references. Use when the user asks for a docs canvas, documentation overview, architecture walkthrough, API reference page, or wants to render structured documentation as an interactive canvas.
Conductor/subagent routing for tasks across multiple harnesses. Assesses task complexity and intent, then dispatches to the appropriate subagent model via a harness-specific adapter. Supports Anthropic (Claude), Cursor, OpenCode, Codex (OpenAI), and Pi adapters. Use via /dynamic-task command or direct invocation from other skills.
Agent evaluation framework. Measure agent performance, identify weaknesses, and track improvement over time. Use when assessing agent quality, comparing approaches, or validating changes.
Interview the user relentlessly about a plan or design until reaching shared understanding, resolving each branch of the decision tree. Use when the user wants to stress-test a plan, get grilled on their design, or mentions "grill me".
| name | debugging-and-error-recovery |
| description | Guides systematic root-cause debugging. Use when tests fail, builds break, behavior does not match expectations, or any unexpected error appears. |
| metadata | {"category":"user-invoked"} |
| disable-model-invocation | true |
Default output: return only the result, blockers, and required evidence. Omit preambles, process narration, repeated context, confidence scores, and follow-up offers. Use at most five bullets unless a required artifact or schema needs more.
Adapted from addyosmani/agent-skills (MIT), commit 82ceff41ed4d3c644e3dcca8a0514390b2911223.
Debugging should be systematic. Preserve evidence, reproduce the failure, localize the cause, fix the root cause, and add a guard against recurrence.
When something unexpected happens:
Make the failure happen reliably. If you cannot reproduce it, capture the conditions under which it appears and instrument the system carefully.
Determine which layer is actually failing:
Create the smallest failing example. Strip unrelated code and data until only the bug remains.
Do not patch the symptom if the failure originates elsewhere.
Add a regression test, alert, or verification step that would catch the same failure next time.
Run the relevant targeted checks, then the broader checks needed to prove the system is healthy again.
| Rationalization | Reality |
|---|---|
| "I already know the fix" | Maybe, but reproducing first prevents wasted iterations. |
| "It is flaky, just rerun it" | Flakiness is a real failure mode that should be understood. |
| "It works on my machine" | Environment differences are often the bug. |
| Excuse | Counter |
|---|---|
| "I know what's wrong, no need to reproduce" | Assumptions about root cause are often wrong. Reproduction confirms the actual problem. |
| "I'll just try random fixes until it works" | Random fixes create more bugs. Systematic triage finds the real cause efficiently. |
| "The error message tells me everything" | Error messages show symptoms, not causes. Root cause analysis prevents recurrence. |
| "I'll fix it and move on, no guard test needed" | Without a guard test, the bug can regress. Guard tests prevent future occurrences. |
| "It works on my machine, must be an environment issue" | Environment differences are often the bug. Reproduce in the failing environment. |
