session-palace-builder
Pass
Audited by Gen Agent Trust Hub on Oct 29, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to use
makefor building, cleaning, and testing the session palace components. These are standard local shell operations for development tasks and do not involve remote downloads or suspicious command strings. - [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and organize untrusted session data, including code snippets and conversation history. While this presents an ingestion surface for indirect prompt injection, the skill lacks high-risk capabilities (such as network exfiltration or direct execution of the ingested data) that would allow for exploitation.
- Ingestion points: Session content categories such as 'Conversations', 'Decisions', 'Code', and 'Research' defined in
SKILL.md. - Boundary markers: Not explicitly defined; the skill relies on the agent's internal management of the palace structure.
- Capability inventory: Local execution of
makecommands for build and test cycles. - Sanitization: No explicit sanitization or validation of the session data is mentioned, as the data is used for context organization rather than as parameters for shell commands.
Audit Metadata