Back to skill
Skillv2.0.1

ClawScan security

multi-search-engine · ClawHub's context-aware review of the artifact, metadata, and declared behavior.

Scanner verdict

BenignMar 12, 2026, 7:22 AM
Verdict
benign
Confidence
high
Model
gpt-5-mini
Summary
The skill is internally coherent: it only composes deterministic search URLs from local templates (no network calls or secrets) and the required files and python3 are proportionate to its stated purpose; a few minor metadata/config inconsistencies are noted but not suspicious.
Guidance
This skill appears safe and does what it says: it generates search URLs locally and won't call out to the network or request secrets. Before installing, you may want to: (1) open resources/engine-catalog.json and confirm the listed engine endpoints/domains match your expectations (some entries use region-specific domains, e.g., bing-int points to a cn.bing.com host), (2) note the small metadata/config inconsistencies (declared config.json requirement not used by the script and version numbers differ) — harmless but worth correcting, (3) run the included smoke tests locally (python3 scripts/build_search_urls.py ...) to confirm behavior, and (4) remember that opening any generated URL will contact third-party search engines and may expose query content to those services, so avoid generating URLs containing sensitive secrets.

Review Dimensions

Purpose & Capability
noteName/description (multi-search URL builder and comparator) match the code and resources: the CLI script + engine catalog are appropriate. Minor mismatch: SKILL.md frontmatter lists config.json as a required config but the main script reads only resources/engine-catalog.json (it does not read config.json). Also some metadata files show different version strings (registry lists v2.0.1 while SKILL.md/metadata.json indicate v2.1.0). These are housekeeping inconsistencies, not functional red flags.
Instruction Scope
okSKILL.md instructs the agent to build and inspect search URLs and to only open pages explicitly. The script composes queries, encodes parameters, reads the local catalog, and prints URLs; it does not perform network access, execute shell commands, read other system files, or exfiltrate data.
Install Mechanism
okNo install spec; the skill is instruction-only with a single Python script. The only runtime dependency is python3 and there are no downloads or archive extraction. This is low-risk and proportionate.
Credentials
okThe skill requests no environment variables or secrets. Required config paths are local JSON/text resources. The script does not access environment variables or credentials. Proportional to purpose.
Persistence & Privilege
okalways:false and no mechanism to persist or modify other skills or system-wide agent settings. The skill does not write files or store tokens; it only prints structured output to stdout.