scientific-writing-outline
Pass
Audited by Gen Agent Trust Hub on Jun 17, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION]: The skill processes user-provided content from
story.mdto generate the outline. This creates a surface for indirect prompt injection, as the input file could theoretically contain instructions to override the agent's behavior. However, the risk is considered safe because the skill's actions are limited to generating markdown documentation and the process includes a multi-pass review loop. - Ingestion points: Reads
story.mdin the current directory. - Boundary markers: None defined for the input content.
- Capability inventory: Writing text files (
outline.md,references.md) and performing web-based citation checks. - Sanitization: No explicit content sanitization, but the structured multi-pass review process provides logical validation.
- [DATA_EXFILTRATION]: The skill performs network operations to verify the existence and validity of academic citations via a subagent. This network activity is restricted to the specific task of citation verification and does not involve accessing or transmitting sensitive system information or user credentials.
Audit Metadata