summarize-signal
Pass
Audited by Gen Agent Trust Hub on Sep 3, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill workflow fetches content from external sources like Substack and arXiv, which creates a potential vector for indirect prompt injection if the fetched content contains malicious instructions designed to influence the model's behavior.
- Ingestion points: Data retrieved from external URLs using the
WebFetchtool as described in Step 1 of the workflow inSKILL.md. - Boundary markers: The instructions lack specific delimiters or instructions to the model to ignore or sequester embedded instructions found within the fetched content.
- Capability inventory: The skill's operations are limited to text analysis and JSON generation; there are no capabilities for file system modification, arbitrary command execution, or network exfiltration identified in the provided file.
- Sanitization: There are no explicit steps provided for the sanitization or validation of the fetched external text before it is summarized or classified.
Audit Metadata