debugging
/debugging
Use when you meet an error, a failing test, unexpected behavior, slowness or a build error, BEFORE proposing a fix. No fix is made before the root cause is found. The main chat and every agent that writes code load it; the reviewer may load it when a root-cause analysis is needed.
- Author
- AlicanKaya192 ↗
- Repository
- Mindgate ↗
- Licence
- MIT
- Last repository change
- 11.11.2026
Assessment and security
- Files in this package
- 1
- Scripts
- 0
- HTML comments
- 1
- URLs in SKILL.md
- 0
No signals detected. Repository changed on 11.11.2026, without references to retired models or APIs.
MIT, open: use and modify while retaining author attribution.
- SKILL.md zawiera komentarze HTML (1). Nie widać ich w podglądzie Markdown, a model je czyta.
What it does and when to use it
Use when you meet an error, a failing test, unexpected behavior, slowness or a build error, BEFORE proposing a fix. No fix is made before the root cause is found. The main chat and every agent that writes code load it; the reviewer may load it when a root-cause analysis is needed.
Original author description
Use when you meet an error, a failing test, unexpected behavior, slowness or a build error, BEFORE proposing a fix. No fix is made before the root cause is found. The main chat and every agent that writes code load it; the reviewer may load it when a root-cause analysis is needed.
How it works
- Basic rule
- 0. Vault first
- 1. Root-cause investigation (before fixing)
- 2. Pattern comparison
- 3. Hypothesis and test
- 4. The fix
- The 3-attempt rule
- Stop signs
- When there really is no root cause
- When done
SKILL.md preview
--- name: debugging description: Use when you meet an error, a failing test, unexpected behavior, slowness or a build error, BEFORE proposing a fix. No fix is made before the root cause is found. The main chat and every agent that writes code load it; the reviewer may load it when a root-cause analysis is needed. --- Method skill (not an agent). Works together with rule #24 (verify first) and §7.3 (problem and solution) of the vault `CLAUDE.md`. ## Basic rule **No fix without a root cause.** Suppressing the symptom (e.g. swallowing the error with `try/except`, adding a magic wait) is not a fix; the problem comes back somewhere else. ## 0. Vault first Search `problems/` for the error message and for similar meaning (`grep`, `search.py`; §7.3, rule #38). If a recorded solution exists, try it first; do not retry in the same conditions one marked ✗ in the "Tried solutions" table. Write the result (✔/✗/~ and why) into the table; if it did not work, look for a new one without deleting the old one. ## 1. Root-cause investigation (before fixing) - **Read the error message to the end:** stack trace, file:line, error code. The answer is often there. - **Reproduce:** Are the steps exact, does it happen every time? If it cannot be reproduced, do not guess; collect data. - **Look at recent changes:** `git diff`, recent commits, new dependency, setting, environment difference. - **In a multi-layer system, measure the boundaries:** At each component boundary (e.g. script → subprocess → file) print the data going in and coming out once. Prove at which layer it breaks, then go down into that layer. - **Trace the data backwards:** Where did the wrong value come from, who called with this value? Go up to the source; fix at the source. ## 2. Pattern comparison - Find similar working code in the same project. List **every** difference between the working one and the broken one (do not say "this can't matter"). - If you are applying a pattern or an example, read the reference from start to end; do not skim. ## 3. Hypothesis and test - Write a single hypothesis: "X is the root cause, because Y." - Test with the **smallest** change; one variable at a time. If it did not hold, do not add a new fix on top; revert the change and form a new hypothesis. - If you do not understand, say "I don't understand X" (#24). Do not pretend. ## 4. The fix - First write a **failing test** that catches the error, or a small reproduction script (`test-writing`). See the test fail. - **One** fix for the root cause. No "while I'm here" improvement or bulk refactor (#23, #25). - Verify (`verification`): the test is green, other tests are not broken, the original symptom is really gone. ## The 3-attempt rule Once three fix attempts have been made, **stop**. If every fix opens a new problem somewhere else, the problem is most likely in the design, not in a single line. Before a fourth attempt, write to the user the situation, what was tried and the architectural question. ## Stop signs If any of these crosses your mind, go back to step 1: - "Let me patch it quickly for now, I'll look later." - "Let me change this and see." - "Let me change a few things at once and run the tests." - "It's probably X." - Writing a list of solutions without tracing the data flow. - Warnings from the user such as "don't guess" or "isn't this working?". ## When there really is no root cause If the problem comes from the environment, timing or an external service, write what was investigated, put in a suitable mitigation (timeout, retry, meaningful error message) and add a record. This is rare; most of the time the investigation was incomplete. ## When done Write the solved problem to `problems/` or update the existing note (`/problem`): symptom (error message verbatim), root cause, valid solution, **all attempted solutions with their results** (including the ones that did not work, rule #38), how to prevent it. ## Idea sources - obra/superpowers `systematic-debugging` (MIT, Jesse Vincent). Read from a downloaded copy; its match with the official repository was not verified. ## Lessons <!-- The central agent adds verified lessons here: general and reusable, without project names or dates. -->
Versions
| Repository | Score | Licence |
|---|---|---|
| This page · AlicanKaya192/Mindgate | 100 | MIT |
| sentimony/skills | 100 | MIT |
| sentimony/skills | 100 | MIT |
| wrongstack/wrongstack | 100 | MIT |
| dxos/dxos | 100 | Apache-2.0 |
| iliaal/ai-skills | 100 | MIT |
| t8jwscc7qy-sketch/foma-skills | 100 | Apache-2.0 |
| arxdsilva/ai-loops | 100 | MIT |
| zts212653/clowder-ai | 100 | MIT |
Database snapshot: 13.11.2026 · package version 92548066864d (SHA-256 digest of the entire package)
Downloads in the last 30 days: 0 · ranking
Report a problem with this skill: harmful content, incorrect descriptions or an author’s removal request.
Similar skills
- 100systematic-debugging
Use when encountering any bug, test failure, or unexpected behavior, before proposing fixes. Implements scientific method for debugging with root cause analysis. - 92systematic-debugging
当遇到任何 bug、测试失败或意外行为时,在提出修复之前必须使用 - 100systematic-debugging
Four-phase debugging framework that ensures root cause investigation before attempting fixes. Never jump to solutions. Use when encountering any bug, test failure, or unexpected behavior, before proposing fixes. - 100systematic-debugging
Use when encountering any bug, test failure, or unexpected behavior. 4-phase root cause investigation — NO fixes without understanding the problem first. - 90root-cause-debugging
Use when behavior is wrong or broke and the cause is unknown, or someone asks what went wrong: a failing test, wrong totals or output, a 500 in production, something that worked before, a flaky run. Not for adding logging, debug settings, explaining traces or tests for new features. - 100systematic-debugging
Use when encountering any bug, test failure, unexpected behavior, or error. Triggers on: "it's broken", "this doesn't work", "I'm getting an error", "fix this bug", any unexpected behavior. Requires finding root cause before attempting fixes. Symptom fixes are failure.