post
Pass
Audited by Gen Agent Trust Hub on Sep 13, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process content from external sources which may be attacker-controlled.
- Ingestion points: As described in Step 1 of SKILL.md, the agent uses browser tools to fetch content from social media URLs (X, LinkedIn, Bluesky, Threads) and their replies, or processes pasted text provided by the user.
- Boundary markers: SKILL.md lacks explicit delimiters or instructions to treat external data as untrusted, increasing the risk that the agent may follow instructions embedded within the posts.
- Capability inventory: The skill as defined in SKILL.md has the capability to perform network read operations through browser tools and generate text drafts; no high-risk capabilities like file system writes, privilege escalation, or arbitrary code execution were identified.
- Sanitization: There is no evidence in SKILL.md of input validation, filtering, or sanitization logic to mitigate the risk of processing malicious instructions embedded in social media data.
Audit Metadata