auto-memory
Pass
Audited by Gen Agent Trust Hub on Apr 5, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [PROMPT_INJECTION]: The skill creates an indirect prompt injection surface by extracting information from user conversations and persisting it to a 'steering' file (.kiro/steering/memory.md) that is automatically loaded in subsequent sessions to influence agent behavior.
- Ingestion points: Live conversation history and user-provided feedback (Phase 1).
- Boundary markers: None specified for the output file format; entries are appended as raw text lines.
- Capability inventory: File system read and write access to the project's configuration directory (Phase 2 and 3).
- Sanitization: No explicit validation or filtering of extracted content is performed before persistence.
- [COMMAND_EXECUTION]: The skill performs local file system operations to manage persistent project state, including reading, writing, and archiving files within the .kiro/ directory structure.
Audit Metadata