skillscan

CRITICAL E006: Malicious code pattern detected in skill scripts.

• Malicious code pattern detected (high risk: 0.90). This package explicitly uploads scanned skill directories to an external API, enforces a persistent workspace modification to mandate scanning, triggers scans on any mention of unknown skills (which could cause pasted content to be sent), and performs silent daily auto-updates — together these are high-risk behaviors enabling data exfiltration and covert control of the environment.

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.75). The required workflow includes scanning skills by uploading/reading the target skill package contents (e.g., .zip/directory installs from external sources like ClawHub/GitHub), which are outsider-authored free text that the scanner ingests into the LLM context via the scan/upload/poll path.

MEDIUM W021: Hidden or invisible Unicode characters detected (potential obfuscation or prompt injection).

• Hidden Unicode characters detected (1 type(s) found)