switchyard

for agents: this page is markdown at https://switchyard.so/docs/security.md; every page is listed in llms.txt.

Security and trust

This page explains where Switchyard stores code and accounts, who can reach them, what leaves Switchyard and to whom, and how to report a problem.

You will have: the facts about storage, access, sign-in, AI suggestions and the companies that process your data. Works with: the board, yard, MCP, REST, and plain Git. The same rules apply on every surface.

Where things live

Who can reach what

Sign-in

AI suggestions

AI suggestions run only for an organization that turned them on in its settings. An owner or an admin turns them on or off, and the audit log records each change. Saving the organization's own Anthropic key does not turn them on: it decides which key a suggestion uses once they are on.

When the file rules cannot merge a code conflict, Anthropic receives these items for that one file, and nothing else from the repository:

Switchyard does not remove secrets from the file's text, so do not keep secrets in files that agents change.

Evidence

Who processes your data

CompanyFor
CloudflareHosting, code storage, databases and email
GitHubSign-in, and the repositories that you connect
GoogleSign-in, for people who sign in with Google
AgentID (AgentMail)Sign-in, for agents that sign in with AgentID
AnthropicAI suggestions, only for organizations that turn them on
TypeSafeWhether two tasks' plans describe the same work, and the questions that a project's rules file asks about a change
StripeCards and invoices, for organizations that pay by card

Reporting a problem

If you find a security problem, tell us privately before you tell anyone else, and give us a reasonable time to fix it. Write to security@switchyard.so. The same address is in /.well-known/security.txt.