mcpbeat Sign in

Socratic Debugging Agent Skill

Question-only debugging: guide users to find root causes themselves.

2k tokens
context cost
the whole folder, loaded on every use
1
files
instructions only
0
copies elsewhere
how many repositories repackaged it
413
stars on the repo
on the repository, not the skill itself

Install

one command, takes just this skill from the repository
npx skills add https://github.com/notque/vexjoy-agent --skill socratic-debugging

The instruction itself

14 sections, as written by the author

Socratic Debugging Skill

Overview

This skill teaches debugging through structured inquiry rather than providing answers, implementing the Socratic Method pattern. You ask questions that lead the user to discover root causes themselves, building lasting investigative skills rather than offering direct solutions.


Instructions

Core Constraints

Guide the user to discover the answer themselves. The user must arrive at the root cause themselves -- giving answers defeats the learning objective. Always read relevant code first using Read/Grep/Glob before formulating questions. Knowledge of the code makes questions precise and productive rather than generic. Follow the 9-phase progression without skipping: jumping to hypothesis questions without establishing symptoms and state leads to guesswork instead of systematic discovery.

Default Workflow Behaviors

Begin with symptoms regardless of how specific the user's description is. Even detailed reports contain unstated assumptions. Ask one question at a time and wait for the response. Multiple questions overwhelm and dilute focus. Mirror the user's terminology (variable names, function names, domain terms) in your questions to reduce friction and show engagement. When the user discovers something, acknowledge it before asking the next question -- silent progression feels like interrogation. After 12 questions without progress toward root cause, trigger escalation to systematic-debugging as a clean handoff.

Question Progression: 9 Phases

Follow these phases in order. Each phase builds evidence for the next.

| Phase | Purpose | Example Questions |

|-------|---------|-------------------|

| 1. Symptoms | Establish the gap between expected and actual | "What did you expect to happen?" / "What actually happened instead?" |

| 2. Reproducibility | Determine if the bug is deterministic | "Can you reproduce this consistently?" / "What conditions trigger it?" |

| 3. Prior Attempts | Focus on fresh approaches | "What have you already tried?" / "What happened when you tried that?" |

| 4. Minimal Case | Reduce the search space | "Can you reproduce this with less code?" / "What is the smallest failing input?" |

| 5. Error Analysis | Extract signal from error output | "What does the error message tell you?" / "Which part of the message is most informative?" |

| 6. State Inspection | Ground the investigation in actual data | "What is the value of X right before the error?" / "What state do you see at that point?" |

| 7. Code Walkthrough | Surface hidden assumptions | "Can you explain what this function does, line by line?" / "What happens at this branch?" |

| 8. Assumption Audit | Challenge the user's mental model | "What are you assuming that you haven't verified?" / "Could that value ever be null here?" |

| 9. Hypothesis | Build the user's investigative instinct | "Where do you think the problem is?" / "Why there specifically?" |

Execution Flow

  • User describes the bug. Read the relevant code silently using Read/Grep/Glob.
  • Ask Phase 1 question. Your first response must be exactly one question with no other text — no preamble, no diagnosis, no code references or examples, no mention of files you read, no announcement of tools used or planned. Even if the bug seems obvious from the code, start with symptoms. Make the question pointed if the answer is likely simple.
  • Listen, acknowledge, ask next question. Format: brief acknowledgment of what they said, then one question advancing toward root cause.
  • Track question count. After 12 questions with no progress toward root cause, trigger escalation offer.
  • When user identifies root cause, confirm their finding and ask what fix they would apply. Let the user propose the fix.

Hints vs. Leading Questions

Questions may contain subtle directional hints. The goal is discovery, not suffering. A good hint directs attention without revealing the answer: asking what a specific value is right before a failure. A bad hint is a leading question that contains the answer: asking whether a specific value could be null. The line: open-ended questions that narrow focus are hints. Leading questions that contain the answer are violations.

Escalation Protocol

After 12 questions without progress, offer cleanly:

> "We have been exploring this for a while. Would you like to switch to direct debugging mode? I can investigate and solve this systematically instead of through questions."

If user accepts, hand off to systematic-debugging with a summary of what has been established:

  • Symptoms identified
  • What has been tried
  • Current hypothesis (if any)
  • Relevant files/lines discovered

Error Handling

User Says "Just Tell Me the Answer"

Cause: User wants direct help, not guided discovery

Solution: Offer to switch modes cleanly. Say: "Would you like to switch to direct debugging mode? I can solve this for you instead." Hand off to systematic-debugging if they accept.

User Is Frustrated

Cause: Too many questions without visible progress, or questions feel generic

Solution: Acknowledge the frustration. Offer escalation. If they want to continue, read more code and ask sharper, more targeted questions. Generic questions indicate you haven't read the code deeply enough.

Bug Is Trivially Obvious From Code

Cause: A typo, missing import, or simple syntax error visible in the source

Solution: Still ask Phase 1, but make the question very pointed -- narrow enough that the user will see the answer immediately. Follow phase progression; pointed questions stay within the Socratic framework.


References

This skill teaches debugging through structured inquiry within these constraints: Maintain the Socratic method by guiding toward answers; always read code before questioning (generic questions signal incomplete code understanding); follow phase progression to build evidence rather than guessing; escalate cleanly at 12 questions without progress rather than continuing to frustrate the user; use the user's terminology to maintain engagement; acknowledge discoveries to keep the dialogue feeling collaborative rather than like interrogation.

Other skills for the same job

different authors, same section of the catalogue
Webapp Testing
by anthropics
vendor ×12

Toolkit for interacting with and testing local web applications using Playwright. Supports verifying frontend functionality, debugging UI behavior, capturing browser screenshots, and viewing browser logs.

6k tokens scripts
Finishing A Development Branch
by ZhanlinCui
×7

Use when implementation is complete, all tests pass, and you need to decide how to integrate the work - guides completion of development work by presenting structured options for merge, PR, or cleanup

1k tokens
Test Driven Development
by w95
×7

Use when implementing any feature or bugfix, before writing implementation code

2k tokens
Systematic Debugging
by ratacat
×7

Use when encountering any bug, test failure, or unexpected behavior, before proposing fixes

10k tokens scripts
Verification Before Completion
by ZhanlinCui
×6

Use when about to claim work is complete, fixed, or passing, before committing or creating PRs - requires running verification commands and confirming output before making any success claims; evidence before assertions always

1k tokens
Backtest Expert
by BaggaT236
×3

Expert guidance for systematic backtesting of trading strategies. Use when developing, testing, stress-testing, or validating quantitative trading strategies. Covers "beating ideas to death" methodology, parameter robustness testing, slippage modeling, bias prevention, and interpreting backtest results. Applicable when user asks about backtesting, strategy validation, robustness testing, avoiding overfitting, or systematic trading development.

15k tokens scripts
Adaptyv
by christophacham
×3

Cloud laboratory platform for automated protein testing and validation. Use when designing proteins and needing experimental validation including binding assays, expression testing, thermostability measurements, enzyme activity assays, or protein sequence optimization. Also use for submitting experiments via API, tracking experiment status, downloading results, optimizing protein sequences for better expression using computational tools (NetSolP, SoluProt, SolubleMPNN, ESM), or managing protein design workflows with wet-lab validation.

16k tokens
Aeon
by christophacham
×3

This skill should be used for time series machine learning tasks including classification, regression, clustering, forecasting, anomaly detection, segmentation, and similarity search. Use when working with temporal data, sequential patterns, or time-indexed observations requiring specialized algorithms beyond standard ML approaches. Particularly suited for univariate and multivariate time series analysis with scikit-learn compatible APIs.

19k tokens

How to use it

Copy the folder

Take notque/socratic-debugging from the repository into ~/.claude/skills for personal use, or into .claude/skills inside a project.

Check the name does not clash

The agent identifies a skill by the name field in its header. Two skills with the same name cannot sit side by side — one of them will be ignored.