handoff
Pass
Audited by Gen Agent Trust Hub on Oct 23, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from the conversation history and user arguments to generate content that directs the behavior of future agent sessions.
- Ingestion points: The entire current conversation history and user-provided arguments (as described in SKILL.md) are used to generate the summary.
- Boundary markers: Absent. There are no instructions to the agent to use delimiters or to treat the ingested conversation content as potentially untrusted data.
- Capability inventory: The skill performs file system writes (saving to the user's OS temporary directory) and generates a 'suggested skills' section, which can be manipulated to recommend malicious tools to subsequent agents.
- Sanitization: The skill contains an explicit instruction to redact sensitive information (API keys, passwords, PII), which helps prevent data exposure but does not mitigate the risk of instruction injection via the conversation text.
Audit Metadata