Skillv1.0.7

ClawScan security

AI Content Brief, Script & Outline Generator — Research Assistant for Video & Image generation · ClawHub's context-aware review of the artifact, metadata, and declared behavior.

Scanner verdict

BenignMar 25, 2026, 6:48 AM

Verdict: benign
Confidence: high
Model: gpt-5-mini
Summary: This is a read-only knowledge/reference skill whose files and runtime instructions are consistent with its stated purpose (planning briefs, prompts and production workflows) and it does not request secrets, install code, or perform network calls.
Guidance: This skill is a local, read-only knowledge base (prompt templates, guides and best-practice markdown) and appears internally consistent with that purpose. Before installing: 1) confirm you trust the publisher/source (homepage is a repo URL; metadata owner id is present) since the content will be read by agents; 2) review the prompt templates for anything you don’t want agents to use (e.g., instructions to generate problematic content or to reproduce copyrighted/real-person likenesses); 3) remember the skill can be invoked autonomously by agents — check any execution skills (ima-image-ai, ima-video-ai, etc.) you pair it with for network access and credential requests; and 4) because this skill requests file_read access to its own references folder, ensure your runtime sandbox enforces filesystem boundaries so the agent cannot read unrelated local files.

Review Dimensions

Purpose & Capability: okName/description describe a knowledge base for multimedia production; the skill is instruction-only and bundles reference markdown files and templates. Required permissions (file_read, a declared knowledge_base config_path) align with providing a local reference library. No unrelated credentials or binaries are requested.
Instruction Scope: okSKILL.md instructs the agent to read documents under the skill's references/ directory (and gives an explicit local path). That matches a knowledge skill: instructions remain advisory and are limited to teaching agents how to form briefs, pick models, and set parameters. There are no instructions to read arbitrary system files, exfiltrate data, or call external endpoints.
Install Mechanism: okNo install spec and no code files are present. This is instruction-only, so nothing will be written to disk by the skill itself beyond the normal skill install process. This is the lowest-risk install pattern.
Credentials: okNo environment variables, no credentials, and no config paths outside the skill's own knowledge_base are required. The declared config_path (~/.openclaw/skills/ima-knowledge-ai/references/) is proportional to a local reference skill.
Persistence & Privilege: okalways is false; the skill is user-invocable and can be called autonomously (platform default). It does not request persistent privileges nor modify other skills' configs. Autonomous invocation is normal and not a standalone red flag here.