debugkit

A structured method for finding the underlying cause of a software bug.

In plain words
What is it for?
Reproduce failures, narrow symptoms, test possible causes, and provide a failing reproduction with a described fix.
Why use it?
It separates diagnosis from fixing by testing reproductions and hypotheses, leaving evidence for someone to review.

Skill for Claude CodeCodex

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add skills/mimukit/skills/debugkit
Any agent
npx skills add mimukit/skills --skill debugkit
Clone the repo
git clone --depth 1 https://github.com/mimukit/skills

Made for: Claude Code, Codex.

Per session 111 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 4,332 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00111 $0.04332
Opus 5 $0.00056 $0.02166
Sonnet 5 $0.00022 $0.00866
Haiku 4.5 $0.00011 $0.00433

Measured yesterday against content hash 6099852aac40, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

debugkit scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured yesterday.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

skills/debugkit/SKILL.md · 241 lines

How it starts

The opening of the file, as written. The whole thing — 241 lines — stays where its author put it; the contents beside it link to each section on GitHub.

debugkit

The skill you reach for when something is broken and nobody knows why. Every other build skill takes intent as its input: a plan, an issue, a diff, a settled decision. debugkit takes a symptom, and its whole job is to turn that symptom into a cause somebody can act on.

It is a single procedure. There are no modes: the ritual runs the same way on every bug, and the branches inside it are decided by what the evidence allows rather than by what the user asked for.

It diagnoses, it never fixes

debugkit mutates the repo freely to learn, and reverts every one of those mutations. Log lines, bisects, config probes, commented-out branches are all fair game, and none of them survive the run. What survives is the cause, a failing reproduction, and a fix described rather than applied.

This boundary is not modesty, it is what makes the report trustworthy. A skill that finds the cause and also lands the cure has already committed to an answer, so what you read afterwards is a rationalization of an edit that already happened. Keeping the diagnosis and the change in separate hands means there is a moment in between where you can disagree.

Edit is present in this skill's tools precisely because probes edit tracked files. That makes the probe ledger load-bearing rather than decorative: it is the only thing standing between a debugging session and somebody's uncommitted work.

When this fires

"Debug this", "why is this failing", "find the root cause", "what's actually causing this", "this test is flaky", "it broke after the upgrade", "it worked yesterday and now it doesn't", "this endpoint got slower", "/debugkit".

Five boundaries, one line each:

  • Not a code review. A review reads a diff for defects in work somebody just wrote. debugkit chases a symptom in code that already ran and misbehaved. A symptom in a diff nobody has executed is a review job, not a debugging one.
  • Not a test plan. Planning how to verify a feature happens before the bug exists. debugkit starts after something has already gone wrong.
  • Not a prototype. A throwaway spike answers an unsettled intent ("would this design hold up?"). debugkit probes to explain an observed failure. The tools look similar; the question is the opposite.
  • Not a feature request. See the intake bar. Behaviour nobody ever built is missing, not broken.
  • Not optimization. A performance regression is in scope: "it used to be fast" has a change to bisect and a clean before-and-after, so the ritual runs unmodified. "Make this faster" is not: it has no cause to find, only a profile to read, and every gate below presumes a working state that stopped working.

Read the full file on GitHub · 241 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. yesterday First seen · 241 lines · 111 tokens per session scan A 6099852aac40

Subscribe to this mod's changes

debugkit is a skill published in the GitHub repository mimukit/skills (1 stars, last pushed yesterday), licensed MIT. It adds 111 tokens to every session and 4,332 once invoked, about $0.0006 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.