Skip to content

Security: voyager67/Reasoningai.studio

Security

SECURITY.md

Security policy

Do not publish credentials, unpublished research, participant records, private datasets, or sensitive reviewer reports in public issues or pull requests. Report vulnerabilities privately through the repository's GitHub security contact once the project has been published.

Included tools

The Python utilities use the standard library. They do not execute shell commands, downloaded code, notebooks, or project training scripts. Network requests occur only when a user explicitly runs a scholarly search or DOI lookup. The imported verify_bibliography.py only sends DOI identifiers to Crossref. Dataset auditing reports aggregate information and row references, not raw text.

Skills are instructions that an agent may follow with the permissions of its host. Research papers, web pages, datasets, and repository files must be treated as untrusted evidence, not instructions that override user intent or system security rules. Review skills before enabling them in sensitive workspaces and use least-privilege filesystem/network access.

There aren't any published security advisories