Skip to content
Back to skills

Diagnose

ASecurity

Use when the user needs failure triage, blocked-lane diagnosis, or root-cause decomposition before the next gate can move.

  • 5 stars
  • 0 votes
  • 0 copies
  • 3 views
  • Added September 23, 2026
ai-agents

Security analysis

A100/100

Pro scans all 2 files and shows the line behind each finding

Scanned October 5, 2026

npx -y skills add JustineDevs/meta-architect --skill diagnose --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Diagnose?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Diagnose
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/justinedevs-diagnose/badge)](https://www.skillsdirectory.com/skills/justinedevs-diagnose)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: diagnose
description: "Use when the user needs failure triage, blocked-lane diagnosis, or root-cause decomposition before the next gate can move."
---

# Diagnose

Use this skill inside Codex when a lane is blocked, a symptom is vague, or the next useful move is to decompose the failure before editing code. `diagnose` is a non-gating helper skill.

## Output

Produce:
- observed symptom
- likely failure slices
- missing evidence or missing reproduction steps
- smallest next probe
- exact next trigger, usually the owning lane or `$maestro`

## Rules

- Decompose the problem before proposing broad fixes.
- Prefer the smallest reproducible boundary that can confirm or kill a hypothesis.
- Keep release-state ownership with the gated lane that is blocked.
- Escalate assumptions clearly when evidence is still incomplete.

## Procedure

1. Restate the observed symptom, expected behavior, scope, and first failing boundary.
2. Build a short hypothesis tree ordered by likelihood and impact.
3. Run the smallest read-only probe that distinguishes the leading hypotheses.
4. Record exact evidence, including versions, routes, inputs, and failure output.
5. Recommend one repair owner and one verification step; do not make unrelated edits.

## Quality bar

- Separate a symptom from a root cause and a contributing condition.
- Never infer success from a stale UI state, cached artifact, or unverified deployment.
- If evidence is insufficient, state the next probe instead of inventing certainty.

## Example

For a plugin skill marked “Error,” inspect the archive root, manifest, skill path, front matter, and server discovery separately before changing the skill text.

Files in this skill

  • SKILL.md837 B
  • agents/openai.yaml267 B

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.