Rechazo resources

Secure the path from prompt to action.

Source-linked security guidance for AI coding agents, tools, prompts, credentials, and permissions.

Choose a route

Three kinds of material. Clear evidence labels.

Use a checklist to make a change, a research note to examine evidence, or updates to follow verified product news.

Checklists for the systems, tools, and actions around AI coding agents.

Source-led notes that separate documented behavior from our interpretation.

Public announcements and release notes, when there is something verified to publish.

Start here

The guides begin with access and action boundaries. The research note shows how to classify the evidence behind those recommendations.

Field guideDeployment checklist

A practical baseline for limiting what coding agents can read, what they can do, and how teams review the result.

For security and engineering teams

Research noteSource synthesis

A source-based map of where untrusted content, agent permissions, tool calls, and security records meet.

For security architects and engineering leads