skill-learning
Pass
Audited by Gen Agent Trust Hub on Mar 14, 2026
Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection (Category 8) because it processes untrusted data from external sources and uses it to modify local files.
- Ingestion points: The skill ingests data from external URLs via
WebFetch(Phase 1a) and from local file systems viaRead(Phase 1b). - Boundary markers: The instructions do not include specific delimiters or directives for the agent to ignore embedded instructions or commands within the source material being analyzed.
- Capability inventory: The agent has the capability to modify the local filesystem using
EditandWritetools (Phase 6a, 6b). - Sanitization: There is no evidence of sanitization or filtering of the extracted content to remove potentially malicious instructions before they are presented to the user for approval.
- [EXTERNAL_DOWNLOADS]: The skill fetches external content to facilitate its learning process.
- It utilizes the
WebFetchtool to retrieve data from user-provided URLs. - It employs
r.jina.ai(a well-known service for LLM-friendly content extraction) as a fallback mechanism for scraping websites.
Audit Metadata