handoff
Pass
Audited by Gen Agent Trust Hub on Sep 29, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill facilitates the execution of local project verification scripts including
pnpm test,pnpm build, andpnpm lintto ensure code quality before a session ends. - [INDIRECT_PROMPT_INJECTION]: The skill aggregates session metadata and project logs to generate a handoff message for future agents, which represents an attack surface for instruction propagation.
- Ingestion points: Processes output from developer tools and reads local documentation files such as
.claude/project-diary.mdandAGENT-LEARNINGS.md. - Boundary markers: The generated handoff prompt uses a specific template but lacks robust delimiters to isolate session-specific data from the agent instructions.
- Capability inventory: The skill interacts with the local file system for logging and executes shell commands for project verification.
- Sanitization: No explicit sanitization or filtering is performed on session summaries before they are embedded into the handoff prompt.
Audit Metadata