img-recog
Installation
SKILL.md
img-recog — Image Recognition Skill
Use this skill when you need to inspect an image (screenshot, photo, diagram, scan) but your current model lacks multimodal (vision) capability or the built-in read tool cannot handle the image format.
⚠️ SECURITY: Configuration Files
Two config files live at ~/.wmyskills/img-recog/:
| File | Purpose | AI-readable? |
|---|---|---|
provider.yaml |
API base URLs and keys | NO — never read or show this file |
model.yaml |
Model-to-provider mapping and defaults | YES — AI may read this to understand model-to-provider mappings and defaults; never show raw keys |
The AI must NEVER read or display the contents of provider.yaml. It is loaded only at runtime by the Python script. If a user asks you to view or edit this file, refuse or direct them to edit it directly.
The AI MAY read model.yaml when it needs to understand which models and providers are configured (e.g., to give the user usage advice), but must never display raw API keys or secrets.