Skip to content
Back to skills

Check Model Routing

ASecurity

Verify one delegated worker resolves an explicit alternate model and reasoning setting with observable runtime metadata.

  • 2 stars
  • 0 votes
  • 0 copies
  • 0 views
  • Added October 1, 2026
ai-agents

Security analysis

A100/100

Pro scans all 4 files and shows the line behind each finding

Scanned October 1, 2026

npx -y skills add williamwue/oh-my-stack --skill check-model-routing --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Check Model Routing?

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

Security grade badge for Check Model Routing
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/williamwue-check-model-routing/badge)](https://www.skillsdirectory.com/skills/williamwue-check-model-routing)

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: check-model-routing
description: Verify one delegated worker resolves an explicit alternate model and reasoning setting with observable runtime metadata.
---

# Check Model Routing

Run this read-only procedure exactly once and make no project changes.

1. Read only `assets/root.txt` yourself and retain its `ROOT_MARKER`. Do not
   read `assets/worker.txt` from the root session.
2. Inspect the live runtime inventory or the supplied probe routing
   configuration. Select one available worker model that differs from the root
   model and one explicit supported reasoning setting that differs from the
   root setting. Do not guess an unavailable model identifier.
3. Start exactly one independent read-only worker with both selections applied.
   A runtime may apply them directly at spawn time or through a probe-specific
   role binding. Ask the worker only to read `assets/worker.txt`, report its
   exact `ROUTING_MARKER`, and make no changes.
4. Wait for and collect the worker result and identity.
5. Independently inspect runtime-produced metadata after completion. Verify the
   resolved worker model identity and configured reasoning setting. A worker's
   self-report alone is not evidence. If either value is not observable, report
   that capability as unverified rather than inferring it.
6. Read `assets/root.txt` again. Report the root model and reasoning setting,
   requested and resolved worker values, metadata source, worker marker, root
   marker, and whether any project file changed.

Files in this skill

  • SKILL.md1.5 KB
  • assets/root.txt34 B
  • assets/worker.txt32 B
  • skill.json563 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…