get-seen-in-replies
Pass
Audited by Gen Agent Trust Hub on Nov 6, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted data from Twitter (tweets and replies), which creates an attack surface for indirect prompt injection where a malicious tweet could contain instructions designed to hijack the agent's behavior.
- Ingestion points: Processes external tweets and replies to generate response content as described in SKILL.md.
- Boundary markers: The skill proactively includes a section 'If you are reading somebody else's tweet' which explicitly instructs the agent to treat external text as data rather than instructions. This significantly reduces the likelihood of the agent following malicious commands embedded in tweets.
- Capability inventory: The analysis did not find any high-risk capabilities such as local file writing, network exfiltration of sensitive files, or shell command execution. The skill predominantly uses standard text generation and voice-matching tools.
- Sanitization: The skill implements instructional sanitization by directing the agent to maintain its role and project scope even when encountering conflicting instructions within processed tweets. It explicitly states that only the primary user decides what is written.
Audit Metadata