Clawra Selfie

Edit Clawra's reference image with Grok Imagine (xAI Aurora) and send selfies to messaging channels via OpenClaw

Before installing: (1) note the SKILL.md requires FAL_KEY and OPENCLAW_GATEWAY_TOKEN but the registry metadata does not list them — ask the publisher to update metadata to declare required secrets. (2) Understand what those secrets allow: FAL_KEY allows billed calls to fal.ai image-editing APIs; OPENCLAW_GATEWAY_TOKEN authorizes the OpenClaw gateway to send messages to channels (so treat it like a messaging credential). Only provide a minimal-scoped token and avoid sharing it. (3) Confirm the OpenClaw gateway the skill will hit is local/trusted (the docs reference http://localhost:18789); if your gateway is reachable remotely, the risk increases. (4) Consider policy/privacy issues: the skill generates realistic selfies (faces), which may have platform or legal restrictions. (5) Because this is an instruction-only skill with no provenance (source/homepage unknown), prefer running it in an isolated/testing environment first and request the publisher to: declare required env vars in registry metadata, provide a trusted source/homepage, and document token scopes. If you cannot verify those items, proceed cautiously or avoid installing.

Type: OpenClaw Skill Name: skill-10 Version: 0.1.0 The skill facilitates AI image editing and distribution but contains a shell injection vulnerability in its TypeScript implementation within SKILL.md. Specifically, the use of `child_process.exec` with unsanitized variables (`channel`, `messageCaption`) allows for arbitrary command execution if a user provides crafted input. While the behavior aligns with the stated purpose of using the fal.ai API (fal.run) and OpenClaw CLI, the lack of input sanitization in a high-privilege environment (requiring Bash and network access) poses a significant security risk.