share-memory
Pass
Audited by Gen Agent Trust Hub on Jun 12, 2026
Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [SAFE]: The skill performs state management using local Markdown files within the project directory. It does not possess network access or external dependencies beyond the initial installation.
- [COMMAND_EXECUTION]: Employs standard shell commands for project initialization, file permission management, and optional Git versioning. These operations are scoped to the project environment and are necessary for the skill's core functionality.
- [DATA_EXFILTRATION]: Provides a security mechanism via
check_memory.shthat scans for sensitive strings (e.g., API keys, passwords, private keys) using regular expressions. This acts as a defensive measure to prevent the accidental leakage of credentials into shared memory logs. - [PROMPT_INJECTION]: The protocol explicitly defines that user instructions take precedence over any information stored in memory. This ensures that the agent remains under the user's direct control and prevents the shared memory from being used to inject unauthorized instructions or override safety guardrails.
Audit Metadata