agentleFS
Sign inStart free
alexderzalexderz/agentic-toolbox/skills/debug/SKILL.md

debug

Skill5 starsChanged 46 days agoMITView on GitHub

use this when a test fails, a bug appears, or behavior is unexpected — find root cause before any fix. default debugging skill. do not use to skip TDD or to land past an open blocker.

192 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use add-discord, add-slack and 10 more.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • MicrosoftSkill

    add-discord

    Add Discord bot channel integration via Chat SDK.

    Also in 11 projects that use this

  • MicrosoftSkill

    add-slack

    Add Slack channel integration via Chat SDK.

    Also in 11 projects that use this

  • +10 more from microsoft

The file

We split this skill into the rules it gives an agent

6 rules. 0 are common, 5 are distinctive, and 1 is very specific to this project.

  • 0

    Common

    Most of the 175 edited versions we read keep these.

  • 5

    Distinctive

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Test failures, unexpected behavior, build/integration breaks, performance.

    line 18
  2. Distinctive:

    Read the full error and stack.

    line 24
  3. Distinctive:

    Check recent changes (diff, deps, config, env).

    line 26
  4. Very specific:

    Verify (verify-before-done).

    line 42
  5. Distinctive:

    After a confirmed cause, add a guard if the layer can lie (validation at the boundary that failed).

    line 48
  6. Distinctive:

    Remove temporary instrumentation before land.

    line 49

“Common” is measured: kept by at least 6 in 10 of the 175 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Debug (default)

Rewrite of obra/superpowers systematic-debugging @ b36e0829. Compress, not a paste. Alternatives in this repo: debug-pocock (tight loop first), debug-anthropic (short report). Load one.

Pairs with tdd (failing test for the cause) and verify-before-done (prove the fix). No scripts/.

Iron law

No fix without root cause. Phase 1 first. Symptom patches are failure.

When

Test failures, unexpected behavior, build/integration breaks, performance. Especially under time pressure or after a failed “quick fix.”

Phases (in order)

1. Root cause

  • Read the full error and stack. Note files, lines, codes.
  • Reproduce. If you cannot, gather data — do not guess.
  • Check recent changes (diff, deps, config, env).
  • Multi-component: log enter/exit at each boundary, run once, see where it breaks, then go there.
  • Deep stack: trace the bad value up to its source. Fix the source.

2. Pattern

Find working similar code. Read the reference completely. List every difference. Name dependencies and assumptions.

3. Hypothesis

One hypothesis: “X is the root cause because Y.” Smallest test of that one variable. Worked → phase 4. Failed → new hypothesis, do not stack fixes. If you do not know, say so.

4. Fix

  1. Failing test for this cause (tdd).
  2. One change. No “while I’m here.”
  3. Verify (verify-before-done). Full relevant suite, not vibes.
  4. Failed fix: if fewer than 3 attempts, back to phase 1. If 3+ failed: stop. Question architecture with the operator. Do not attempt fix #4 alone.

Always

  • One variable at a time.
  • After a confirmed cause, add a guard if the layer can lie (validation at the boundary that failed).
  • Remove temporary instrumentation before land.

Ask first

  • Fourth fix attempt or an architectural change after 3 failures.
  • Production-only instrumentation.

Never

  • “Quick fix now, investigate later.”
  • Several untested changes in one run.
  • Skip the failing test.
  • Land past an open blocker because debugging “unblocked” you in chat.

Red flags

  • “It’s probably X”
  • “Just try changing this”
  • Each fix reveals a new symptom elsewhere
  • Lists of fixes with no investigation

Upstream

obra/superpowers skills/systematic-debugging @ b36e0829c6d0140e93cfef2ca599b1b07d4a7797 (MIT). See SOURCES.md.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

Check your setup, free