alterlab-yok-akademik
Pass
Audited by Gen Agent Trust Hub on Sep 25, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDATA_EXFILTRATIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests data from the external akademik.yok.gov.tr portal to display academic profiles. \n
- Ingestion points: Data is fetched via scripts/yok_akademik.py using HTTP GET/POST requests to the government portal. \n
- Boundary markers: SKILL.md explicitly instructs the agent to treat records as 'authoritative-but-scraped' and confirm matches with the user. \n
- Capability inventory: The skill uses requests and urllib for network reads and BeautifulSoup for parsing; it lacks file-write or execution capabilities on the ingested data. \n
- Sanitization: The _visible_text function in scripts/yok_akademik.py strips HTML tags and scripts, reducing the risk of processing malicious markup. \n- [DATA_EXFILTRATION]: The skill performs network operations to fetch data from a government portal. \n
- Evidence: Connects to akademik.yok.gov.tr, which is a well-known service for Turkish academic data. No sensitive local data is accessed or transmitted. \n- [COMMAND_EXECUTION]: The skill executes a local Python script for its core functionality. \n
- Evidence: SKILL.md uses uv run python to execute scripts/yok_akademik.py. The script is bundled with the skill and performs scoped scraping tasks as described.
Audit Metadata