ccf-literature-monitor
Pass
Audited by Gen Agent Trust Hub on Oct 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external content from academic databases which represents a standard ingestion surface for untrusted data.\n
- Ingestion points: Paper titles, abstracts, and metadata are retrieved from external services like the arXiv API and OpenReview as defined in
references/monitoring-workflow.md.\n - Boundary markers: The skill utilizes specific evaluation axes (problem, mechanism, evidence) and a rigid report template in
references/report-template.mdto structure the output and mitigate the impact of potentially malicious content within abstracts.\n - Capability inventory: The skill is restricted to research analysis and documentation; it does not possess capabilities for arbitrary command execution, sensitive file writes, or network exfiltration of local credentials.\n
- Sanitization: A key security control is documented in the setup phase, requiring the transformation of private research ideas into public-safe query terms before any external network requests are performed.
Audit Metadata