c-memory
Pass
Audited by Gen Agent Trust Hub on Mar 9, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: Uses standard shell commands (cat, ls, echo) to manipulate local files in ~/.claude/memory/ and invokes obsidian-cli for synchronization.
- [PROMPT_INJECTION]: Identifies a surface for indirect prompt injection via the ingestion of untrusted memory data.
- Ingestion points: Reads from ~/.claude/memory/MEMORY.md and obsidian-cli search results.
- Boundary markers: None detected for the interpolated memory content.
- Capability inventory: File reads, writes, and subprocess execution via shell commands.
- Sanitization: No explicit sanitization or validation of the stored memory content before it is read back into the agent context.
Audit Metadata