agent-memory-mcp
Fail
Audited by Gen Agent Trust Hub on Feb 28, 2026
Risk Level: HIGHEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill requires the user to clone an external codebase from 'https://github.com/webzler/agentMemory.git' during the setup process. This repository belongs to an unverified third party, which introduces the risk of downloading malicious or unvetted scripts into the agent's environment.- [REMOTE_CODE_EXECUTION]: The skill facilitates the execution of code downloaded from an external, unverified source. By running 'npm run compile' and 'npm run start-server' on the cloned project, the agent executes scripts whose behavior is not defined or controlled by the skill's own distribution.- [COMMAND_EXECUTION]: The installation and execution steps involve several shell commands, including 'npm install' and 'npm run compile', which can trigger arbitrary script execution via lifecycle hooks in the external repository's package configuration.- [PROMPT_INJECTION]: The skill exhibits a vulnerability surface for indirect prompt injection.
- Ingestion points: Untrusted data enters the context through 'memory_write' content and by scanning the file system at the provided absolute path.
- Boundary markers: There are no explicit delimiters or instructions to ignore embedded commands within the memory retrieval logic.
- Capability inventory: The skill possesses capabilities for persistent file system access and hosting a local web dashboard via Node.js.
- Sanitization: The documentation provides no evidence of sanitization or validation of the content stored in or retrieved from the memory bank.
Recommendations
- AI detected serious security threats
Audit Metadata