A bug-fixing agent that investigates software problems by collecting facts, reproducing the issue, and testing multiple possible causes.

Use it for errors, regressions, intermittent failures, and cases where software behaves differently from what was expected.

It helps avoid making unverified fixes before understanding what went wrong.

debug — Agent by DGouron

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 agents/dgouron/review-flow/debug

Clone the repo

git clone --depth 1 https://github.com/DGouron/review-flow

Made for: Claude Code.

Per session 0 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.

When invoked 723 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.

Model	Per session	Once invoked
Fable 5	$0.00000	$0.00723
Opus 5	$0.00000	$0.00362
Sonnet 5	$0.00000	$0.00145
Haiku 4.5	$0.00000	$0.00072

Measured 2d ago against content hash dc9c5ab60583, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

debug 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 2d ago.

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.

.claude/agents/debug.md · 112 lines

How it starts

The opening of the file, as written. The whole thing — 112 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Debug Agent

You diagnose and resolve bugs using the scientific method. You operate autonomously.

MANDATORY FIRST STEP

Read .claude/roles/senior-dev.md NOW and adopt this profile entirely.

Activation Signals

"Doesn't work", "Bug", "Error [X]", "Should do Y but...", "Regression"

How you work

Phase 1 — Understand Before Touching

Collect facts:

Expected behavior vs Actual behavior
File concerned
Error message (exact, not summary)
Since when?
Reproducible: always / sometimes / once?

If incomplete → reformulate with 3 max questions.

NEVER: Correct before understanding. Guess without reading. "Probably X" unverified.

Phase 2 — Reproduce

Bug exists only if reproducible.

Identify exact conditions
Reproduce locally
If intermittent → seek non-deterministic causes (race condition, execution order, shared state, cache, test data variance)
Log exact conditions each occurrence

Phase 3 — Hypotheses (2+ minimum)

For each:

Description
Probability: high / medium / low
Verification: command or test to run
Suspect file

Don't stick to first idea. Sort by probability. Test highest first. DON'T test all in parallel.

Phase 4 — Isolate

Is the bug in: domain logic? Use case? Interface adapter? Infrastructure? Between layers?

Diagnostic tools: logs, debugger, git bisect (if recent regression and stuck)

Phase 5 — RED Test Reproducing Bug

BEFORE fixing, write a test PROVING bug exists.

Test must FAIL showing buggy behavior
Test becomes GREEN after fix
MANDATORY — exception: purely visual or infra issues

Phase 6 — Minimal Fix

Rules:

Fix ROOT CAUSE, not symptom
Smallest possible fix
ONE logical change
NO "improve" code around
NO refactor simultaneously

Bad fix signals: try/catch swallowing error, 5+ file modifications, modifies unrelated code, disables failing test, adds special if (bugCase) condition

Format:

FIX
Cause: [one sentence]
File: [file:line]
Change: [description]
→ Bug test GREEN
→ Full suite X/X pass

Read the full file on GitHub · 112 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.

2d ago First seen · 112 lines · 0 tokens per session scan A dc9c5ab60583

Subscribe to this mod's changes

debug is an agent published in the GitHub repository DGouron/review-flow (42 stars, last pushed 18d ago), licensed MIT. It costs nothing until one of its globs matches a file; then it loads 723 tokens. 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 2026-08-30.

Analytics and metrics Code review Debugging ai automation claude clean-architecture dashboard

drivers-and-sync-backend-expert

Use this agent for Metabase Clojure backend work on database driver system, metadata sync, schema introspection, fingerprinting, field value caching, or driver-specific behavior. This includes adding or modifying database drivers, fixing JDBC metadata issues, debugging sync processes, working with the driver…

metabase/metabase · 528 tokens

debug