grad-paragraph
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process data from external files such as
outline/subsection_briefs.jsonl,outline/evidence_drafts.jsonl, andcitations/ref.bib. While this represents an attack surface where malicious instructions could be embedded in the data, the skill lacks privileged capabilities (such as network access or shell execution) that could be exploited. The processing of these files is the primary intended purpose of the skill. - Ingestion points: Files
outline/subsection_briefs.jsonl,outline/evidence_drafts.jsonl,outline/evidence_bindings.jsonl, andcitations/ref.bib(referenced in SKILL.md). - Boundary markers: The instructions specify a clear 4-line planning phase and specific prose constraints, which act as a structural boundary, though no explicit "ignore" delimiters are present.
- Capability inventory: No subprocess calls, file-write operations, or network operations are present in the skill instructions.
- Sanitization: No specific sanitization or escaping logic is described for the input data.
Audit Metadata