Skillv1.0.0

ClawScan security

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

Scanner verdict

BenignApr 11, 2026, 2:53 AM

Verdict: benign
Confidence: medium
Model: gpt-5-mini
Summary: The skill appears to do what it says (a local CORS scanner that issues test requests and grades responses) and does not request unrelated credentials or installs, but the shipped script was truncated in the review input so there's some remaining uncertainty until the full source is inspected.
Guidance: This skill is internally consistent with being a CORS misconfiguration scanner: it only needs to make HTTP(S) requests and inspect headers. Before running it, consider these practical points: (1) the included script disables TLS verification (ssl.CERT_NONE) and turns off hostname checks — this is common in scanning tools to avoid failing on self-signed certs but means the scanner will accept invalid certificates; (2) the scanner will send crafted Origin headers and may trigger server-side behavior—only scan endpoints you own or have permission to test to avoid legal issues; (3) I reviewed most of the script, but the provided source was truncated in the listing — inspect the full scripts/cors_scan.py locally to confirm there are no calls that log or transmit results to third-party services before executing; (4) run the tool in an isolated environment (or sandbox) if you want extra caution. If you can provide the complete scripts/cors_scan.py contents I can raise my confidence to high.
Findings: [no_regex_findings] expected: Static pre-scan found no patterns of concern. For a pure network scanner this is expected; absence of findings does not guarantee there is no unexpected behavior in the portions of the file not shown.

Review Dimensions

Purpose & Capability: okName/description align with the included script: the code issues HTTP(S) requests with crafted Origin headers and inspects CORS response headers to detect reflection, wildcards, null origin, preflight issues, private-network access, etc. No unrelated credentials, binaries, or services are requested.
Instruction Scope: okSKILL.md simply instructs running the included Python script against target URLs and describes expected outputs/formats. The instructions do not ask the agent to read local files, environment variables, or send results to external endpoints. Note: the script performs network requests to arbitrary targets (expected for a scanner).
Install Mechanism: okNo install spec is present and the tool uses only Python standard library (urllib, ssl, argparse). This is proportionate for a lightweight scanner. The package is delivered as source (scripts/cors_scan.py) so no external downloads during install.
Credentials: okThe skill declares no required environment variables, credentials, or config paths. The script does not appear to reference secrets. This is appropriate for a network-scanning utility.
Persistence & Privilege: okalways:false and no special privileges or modifications to other skills/systems are requested. The skill does not request permanent presence or elevated agent configuration.