memory-sync
Pass
Audited by Gen Agent Trust Hub on Aug 15, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill establishes a mechanism for indirect prompt injection by persisting session data into a long-term memory file that influences future agent behavior.
- Ingestion points: The agent is instructed to read
~/.ai-software-agency/memory/<agent-name>/MEMORY.mdat the start of every session to load context and identity. - Boundary markers: The instructions do not define specific delimiters or warnings to ignore potentially malicious embedded instructions within the retrieved memory content.
- Capability inventory: The agent performs file read and write operations within the
~/.ai-software-agency/directory structure, which includes core project documentation (PRD, Design, Tests) and audit logs. - Sanitization: There is no provided logic for sanitizing, validating, or filtering session content before it is recorded in the Decision Log (Section 4) or Learned Patterns (Section 5), allowing arbitrary strings to be persisted.
Audit Metadata