Skip to content
Skills Library
Programming · Analysis

debug-agent

/debug-agent

Diagnose and fix software bugs through scoped reproduction, causal investigation, focused repairs and regression verification. Use for debugging requests, especially pause, checkpoint/resume and hook failures; skip feature development and conceptual questions without a fault to investigate.

Repository
debug-agent ↗
Licence
MIT
Last repository change
30.10.2026

Assessment and security

91/ 100Class: good
Format25 / 25
Frontmatter with name and description, valid name, description up to 1,024 characters, name matching the folder.
Discoverability25 / 25
Description explains when to use the skill, has a reasonable length and uses third person.
Content26 / 35
At least 150 words and at most 500 lines, headings, step lists, examples or code.
Hygiene15 / 15
No placeholders such as TODO, no Windows paths, with a licence.
Riskno signals detected in the scan
Files in this package
8
Scripts
0
HTML comments
0
URLs in SKILL.md
0
Deprecation

No signals detected. Repository changed on 30.10.2026, without references to retired models or APIs.

Licence

MIT, open: use and modify while retaining author attribution.

What it does and when to use it

Diagnose and fix software bugs through scoped reproduction, causal investigation, focused repairs and regression verification. Use for debugging requests, especially pause, checkpoint/resume and hook failures; skip feature development and conceptual questions without a fault to investigate.

Original author description

Diagnose and fix software bugs through scoped reproduction, causal investigation, focused repairs and regression verification. Use for debugging requests, especially pause, checkpoint/resume and hook failures; skip feature development and conceptual questions without a fault to investigate.

How it works

SKILL.md has no level-two headings · 32 lines, 218 words

SKILL.md preview
---
name: debug-agent
description: Diagnose and fix software bugs through scoped reproduction, causal investigation, focused repairs and regression verification. Use for debugging requests, especially pause, checkpoint/resume and hook failures; skip feature development and conceptual questions without a fault to investigate.
license: MIT
metadata:
  version: "0.1.4"
---

# Debug Agent

Use the [runner](AGENT.md) and [playbook](PLAYBOOK.md) for the current bug.
Resolve those files, [configuration](agent.json), [toolbox](toolbox/index.json)
and [Codex adapter](adapters/codex.md) relative to this skill directory.

On invocation, discover current target evidence and establish expected versus
actual behavior. Reproduce safely, freeze source-based outcomes, diagnose the
cause, apply necessary authorized reversible repairs, and verify the original
failure plus meaningful regressions. Preserve current user edits and ownership.

The mode is manual in the invoking chat. Verify required host capabilities;
the portable configuration's draft status is not a reason for a setup interview
when the current request and tools settle the case. Ask only consequential
missing questions and reuse current authorization.

Store case outputs and reviews in the target's existing approved workspace or
an explicitly selected case folder, never this installed skill directory. Reuse
authoritative checkpoints/queues. Saved files grant no permission. Read the installed
checkpoint contract before applying a recovery procedure: task-checkpoint 0.8 has
no persistent quota pause or mandatory recovery review. Explicit human holds,
native permissions and existing ownership remain controlling. See the
[version compatibility guide](toolbox/checkpoint-compatibility.md).

For detailed examples, consult [cases](toolbox/cases.json). For each actual case,
adapt the [review template](toolbox/run-review-template.json); the package's
construction acceptance does not replace bug-specific checks.

This skill uses the current chat's model/tools. It starts no background process
and does not bypass sandbox permissions, hook trust or resource stops.
Files in this package (8)
  • .gitignore 1 KB
  • AGENT.md 2 KB
  • LICENSE 1 KB
  • PLAYBOOK.md 14 KB
  • README.md 4 KB
  • SKILL.md 2 KB
  • VALIDATION.md 2 KB
  • agent.json 2 KB

Versions

RepositoryScoreLicence
This page · HarryBurd33/debug-agent91MIT
mlflow/skills90Apache-2.0

Database snapshot: 13.11.2026 · package version 682c030c9d41 (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

  • 81shadow-debug
    Diagnose and fix a reported failure using a reproducible symptom, a focused cause investigation, and regression verification. Use for broken behavior, errors, crashes, or failing checks.
  • 85debug-task
    Reproduce and diagnose a specific software defect, then apply and verify a repair only when authorized. Use for executable debugging rather than an analysis-only explanation or general repository survey.
  • 90sbp-debug-investigation
    Systematically investigate, reproduce, and fix software bugs — frame the symptoms, reliably reproduce, isolate the root cause, write a regression test, then fix. Use when chasing a bug, especially in mission-critical systems where "works on my machine" is not an acceptable resolution.
  • 90diagnosing-bugs
    Diagnose bugs and performance regressions through reproduction, ranked hypotheses and causal checks. Use for failures, slow behavior or "debug this"; an explicitly selected host debug command keeps its route.
  • 86bugfix
    Diagnose and repair an existing software defect through bug report, reproduction, fail-first regression test, causal fix, and focused review. Use for reported incorrect, regressed, or intermittent existing behavior; this workflow owns its regression check, so do not also invoke…
  • 100debugging
    Systematically diagnose and fix software bugs by analyzing error messages, stack traces, logs, and runtime behavior across multiple languages. Use when the user requests debugging or provides relevant inputs for this workflow.

Feedback and reports

Your message goes to Bartosz Sokoliński. An email address is needed only for a reply. See the privacy policy.