Back to skill
Skillv1.1.1
ClawScan security
Care Taker · ClawHub's context-aware review of the artifact, metadata, and declared behavior.
Scanner verdict
BenignApr 1, 2026, 1:44 AM
- Verdict
- benign
- Confidence
- high
- Model
- gpt-5-mini
- Summary
- The skill's requirements and runtime instructions match its virtual-pet purpose and it asks for no unrelated credentials or installs — nothing appears disproportionate.
- Guidance
- This skill appears coherent and only uses the animalhouse.ai API. Before installing, confirm you trust animalhouse.ai (homepage/repo), store the returned ah_ token securely (it is shown once), and avoid copying that token into public places. Note that runtime behavior will follow the platform's `next_steps` responses — if you plan to let the agent act autonomously, understand the agent may make further API calls to the service based on those responses. Because this is instruction-only, no code is installed on your system by the skill itself.
Review Dimensions
- Purpose & Capability
- okName/description (virtual caretaker for animalhouse.ai) align with the runtime instructions, which only call the animalhouse.ai HTTP API (register, adopt, status, care). The skill declares no extra binaries, env vars, or config paths that would be out of scope.
- Instruction Scope
- noteSKILL.md contains concrete curl examples to register, adopt, check status, and perform care actions — all scoped to animalhouse.ai endpoints. It instructs the agent to follow the service-provided `next_steps` in responses; this is expected for a guided API but gives the remote service some control over suggested agent actions, so be aware that runtime behavior will depend on server replies.
- Install Mechanism
- okInstruction-only skill with no install spec and no code files — nothing is written to disk or downloaded during install.
- Credentials
- okThe skill requests no environment variables or credentials up front. It expects an API token obtained via the platform's register endpoint (described in the docs), which is proportionate to the stated purpose.
- Persistence & Privilege
- okThe skill is not forced-always, is user-invocable, and uses the platform default allowing autonomous invocation. That default is expected; the skill does not request persistent system-wide privileges or modify other skills.