inner-life-memory

Pass

Audited by Gen Agent Trust Hub on Mar 6, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill requires the 'jq' binary and directs the user to execute an initialization script ('init.sh') from a prerequisite skill. These operations are used for JSON processing and environment setup within the vendor's ecosystem.
  • [EXTERNAL_DOWNLOADS]: Instructions include the installation of prerequisite components using 'clawhub', a third-party installation tool, to fetch necessary resources from the vendor's repositories.
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection due to its core memory extraction logic.
  • Ingestion points: The skill reads untrusted user data from the 'memory/daily-notes/' and 'memory/diary/' directories.
  • Boundary markers: No clear delimiters or instructions to ignore embedded behavioral overrides are provided for the reflection phase.
  • Capability inventory: Processed data is written to critical state files like 'memory/inner-state.json' and 'memory/MEMORY.md' using 'jq' for transformation.
  • Sanitization: The documentation does not specify any validation or sanitization steps to prevent malicious instructions in notes from being interpreted as legitimate memories or principles.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 6, 2026, 03:43 AM