write-like-meng-on-x
Pass
Audited by Gen Agent Trust Hub on Oct 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes social media posts which may contain instructions intended to influence the agent.
- Ingestion points: Data is ingested from external batches into
references/tweet-corpus.jsonlvia a bundled Node.js script. - Boundary markers: Instructions explicitly direct the agent to treat posts as "voice evidence" and avoid splicing text together, reducing the likelihood of executing embedded instructions.
- Capability inventory: The skill executes local management scripts (
node,python3), performs file search usingrg, and writes to its local corpus. - Sanitization: The ingestion script validates data structure and format, while the prompt provides instructions to prevent the agent from treating the corpus data as commands.
Audit Metadata