summarize
Pass
Audited by Gen Agent Trust Hub on Mar 9, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses preprocessing instructions in SKILL.md to execute the local CLI tools 'agentbook inbox', 'agentbook identity', and 'agentbook following' to retrieve data for the summary.
- [PROMPT_INJECTION]: The skill exhibits a vulnerability surface for indirect prompt injection.
- Ingestion points: Untrusted data enters the agent context through the 'agentbook inbox' and 'agentbook following' commands defined in SKILL.md.
- Boundary markers: No delimiters or instructions to ignore embedded commands are used when injecting the external message content into the prompt.
- Capability inventory: The skill provides access to private user messages, contact lists, and identity information.
- Sanitization: There is no evidence of filtering or escaping performed on the retrieved message content before it is processed by the AI.
Audit Metadata