memory
Pass
Audited by Gen Agent Trust Hub on Jun 19, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill implements a persistent memory mechanism that reads from a local file (
C:\Users\lokik\.codex\MEMORY.md) at the start of every session. This creates a surface for Indirect Prompt Injection because the agent treats the content of this file as authoritative context for the new conversation. - Ingestion points:
C:\Users\lokik\.codex\MEMORY.mdis read into the agent's context during initialization. - Boundary markers: The instructions do not specify any delimiters (e.g., XML tags or triple quotes) to separate the loaded memory content from the system prompt, nor does it include warnings to ignore instructions found within the memory data.
- Capability inventory: The skill has the capability to read and write to the local file system (using the provided absolute path).
- Sanitization: There is no logic provided to sanitize or validate the content being written to or read from the memory file. If a previous conversation contained malicious instructions, those instructions would be re-injected into the prompt in future sessions.
Audit Metadata