Skip to content

Security: ScientFactory/scient-agent

SECURITY.md

Security policy

Reporting a vulnerability

Report suspected vulnerabilities privately through the Scient agent security advisory form. Do not open a public issue for an undisclosed vulnerability or include secrets, credentials, private repositories, or sensitive logs in a report.

Include the affected revision, platform, configuration, reproduction steps, expected impact, and the smallest safe proof available. Clearly say whether the issue also reproduces in official OpenCode. A vulnerability that affects both projects may need coordinated reporting, but ScientFactory remains responsible for evaluating its owned source and release boundary.

Current security boundary

This repository is an OpenCode-derived source foundation for the planned native Scient agent. It is not yet a released native Scient agent.

The inherited agent core can execute shell commands, read and write files, use network tools, and connect to configured providers. Its permission prompts are an interaction and approval mechanism, not a security sandbox. Use an operating system account, container, or virtual machine when stronger isolation is required.

Server mode is opt-in. Any exposed server must use authentication and an appropriately restricted network boundary. Provider and MCP-server behavior remain governed by the services the operator chooses.

ScientFactory security decisions, credentials, sessions, permissions, and scientific project truth must remain separate from upstream service or account state.

There aren't any published security advisories