job-posting-intent
Fail
Audited by Gen Agent Trust Hub on Apr 23, 2026
Risk Level: HIGHCREDENTIALS_UNSAFEREMOTE_CODE_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [CREDENTIALS_UNSAFE]: The scripts
scripts/create_sheet.py,scripts/create_sheet_mcp.py, andscripts/search_jobs.pycontain a hardcoded JSON Web Token (RUBE_TOKEN). This credential provides unauthorized access to the Rube/Composio platform. - [REMOTE_CODE_EXECUTION]: The skill programmatically constructs Python code blocks and transmits them to the
RUBE_REMOTE_WORKBENCHtool for execution on a remote server. - [DATA_EXFILTRATION]: Data extracted from external job postings, including company descriptions and website URLs, is transmitted to third-party endpoints at
api.apify.comandrube.app. - [PROMPT_INJECTION]: The skill processes untrusted job description text from LinkedIn to extract buying intent signals. Ingestion points: LinkedIn job data is retrieved via the Apify API in
scripts/search_jobs.py. Boundary markers: No delimiters or safety instructions are used to separate untrusted text from logic. Capability inventory: The skill can perform network operations and execute remote code. Sanitization: Job description text is processed directly for signal extraction without sanitization.
Recommendations
- AI detected serious security threats
Audit Metadata