memory-bridge
Pass
Audited by Gen Agent Trust Hub on May 19, 2026
Risk Level: SAFE
Full Analysis
- [DATA_EXFILTRATION]: The skill is designed to read markdown files from
~/.claude/projects/*/memory/*.mdto extract session memories. While this involves reading local data, it is the core purpose of the tool and the data is stored in the user's AgentDB instance. - [COMMAND_EXECUTION]: The instructions include a CLI command to run a local script (
node .claude/helpers/auto-memory-hook.mjs) to perform bulk imports. This executes code already present in the user's environment. - [PROMPT_INJECTION]: Processing external content from memory files provides a surface for indirect prompt injection. 1. Ingestion points: Reads memories from
~/.claude/projects/*/memory/*.md(SKILL.md). 2. Boundary markers: No specific delimiters are mentioned to isolate memory content from instructions. 3. Capability inventory: The skill utilizesBashand internalmcp__claude-flowtools. 4. Sanitization: No sanitization of ingested markdown content is described.
Audit Metadata