Skillv0.1.0

ClawScan security

GEB Aesthetics · ClawHub's context-aware review of the artifact, metadata, and declared behavior.

Scanner verdict

BenignFeb 11, 2026, 9:32 AM

Verdict: benign
Confidence: high
Model: gpt-5-mini
Summary: This skill's files, instructions, and requirements are coherent with a creative—GEB-inspired—authoring tool and do not request extra credentials or perform suspicious actions.
Guidance: This package appears to be a coherent creative-authoring skill and carries low obvious risk, but please consider these points before installing or running it: 1) Review the GitHub repo (package.json points to one) to confirm authorship and check for additional files (bin/, templates/) that were referenced but not included here. 2) Inspect the scripts (init.sh, generate.sh, export.sh, verify.sh) — they are currently harmless stubs, but always review any scripts before executing them. 3) The skill's package.json lists specific external models; running the skill may cause the agent to call those model endpoints (which could require credentials or network access). Ensure you’re comfortable with the agent using external models and with any billing/credential implications. 4) If you plan to run the scripts on a sensitive system, run them in an isolated directory or sandbox first. Overall: coherent and low-risk, but perform normal repository and script review hygiene before use.

Review Dimensions

Purpose & Capability: noteThe name, README, SKILL.md, and scripts describe a multi-modal creative-spec tool and the declared package.json metadata (models/agents) aligns with a creative AI skill. Minor inconsistencies: package.json references a bin/ directory and templates/ that are not present in the file manifest, and the registry metadata shows 'source: unknown' even though package.json lists a GitHub repo — these look like packaging/metadata sloppiness rather than malicious intent.
Instruction Scope: okSKILL.md is a high-level specification and runtime guidance for generating multi-layer creative artifacts; it does not instruct the agent to read host system files, secrets, network endpoints, or other skills' configurations. The included scripts are simple project scaffolding, generation, export, and verification stubs that only print messages and create project folders/files.
Install Mechanism: okThere is no install spec (instruction-only skill) and no downloads or extract steps. The shipped shell scripts are small, human-readable, and do not fetch remote code or execute obfuscated commands.
Credentials: okThe skill declares no required environment variables, no credentials, and the SKILL.md does not reference any secrets or external auth. The package.json does list preferred models (openai-proxy/gpt-5.3-codex, kimi-coding/k2p5), which may cause the agent to invoke external model endpoints when running the skill, but requesting model access is proportional to a creative AI skill and no credential variables are demanded.
Persistence & Privilege: okalways is false and model invocation is allowed (the platform default). The skill contains no code that modifies other skills or system-wide settings; its scripts create project directories and files within the working directory only.