linkedin-comment-writing
Pass
Audited by Gen Agent Trust Hub on Sep 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted external content (LinkedIn posts) to assist in drafting replies, creating a potential surface for adversarial instructions.
- Ingestion points: The workflow in
SKILL.mdandworkflows/write-value-comments.md(Step 1) requires the agent to analyze the original post's claim, audience, and context. - Boundary markers: The instructions do not provide explicit delimiters or "ignore embedded instructions" warnings for the external post content being analyzed.
- Capability inventory: The skill's capabilities are restricted solely to text generation; it does not utilize shell commands, subprocesses, file writing, or network operations.
- Sanitization: There are no requirements for the agent to sanitize or filter the content of the external posts before processing.
Audit Metadata