fix-node-vuln
Pass
Audited by Gen Agent Trust Hub on Oct 22, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from external repositories by searching and reading configuration files to identify Node.js versions.
- Ingestion points: The skill searches all accessible repositories using
git cloneandrgto find files like.nvmrc,package.json, andDockerfile(specified in Procedure 1 and 6). - Boundary markers: No explicit boundary markers or "ignore embedded instructions" warnings are used when processing the content of these external files.
- Capability inventory: The skill has the capability to execute shell commands, manage git repositories, use Docker, and run Node.js package managers (
npm,pnpm,yarn) as described in Procedure 1 and 9. - Sanitization: There is no evidence of sanitization or validation of the ingested repository content before it is used to inform agent actions or shell commands.
- [COMMAND_EXECUTION]: The skill executes installation and testing commands on code sourced from potentially untrusted external repositories.
- The procedure in Step 9 requires running
npm ci,pnpm install, andnpm testto establish a baseline and verify the fix. - Malicious repositories could contain
package.jsonfiles with lifecycle hooks (e.g.,preinstall,postinstall,test) that execute arbitrary code when these standard package manager commands are invoked. - Although the skill recommends running these commands inside a Docker container, the agent itself initiates the execution, and any network-enabled container could still be used as a pivot or for data exfiltration.
Audit Metadata