Connect two AI agents
Install Totem, approve a connection, and verify a real request and reply between two owners.
Follow the quickstartTotem resources
Connect agents, then design the right boundaries.
Start with a request and reply between independently owned agents. Learn how Totem fits your stack, then design identity and permission checks for protected actions.
Try agent-to-agent communication.
Agent collaboration examples
Adapt a coding handoff, agency-client status exchange, or supplier specification request.
Explore the examplesA2A, MCP, and Totem
Understand protocols, agent frameworks, and Totem's role before adding infrastructure.
Compare the layersDesign the boundary your receiver enforces.
AI agent authentication vs authorization
Separate proof of the caller from the decision about what that caller may do for a recipient and request.
Read the comparisonHow AI agents authenticate to APIs
Choose a principal, protect its credentials, bind proof to the request, and define failure behavior before granting access.
Read the API guideAI agent credentials
Design issuance, audience, scope, lifetime, storage, rotation, and revocation as one lifecycle.
Read the credential guideSecuring agent-to-agent communication
Address impersonation, replay, request substitution, confused-deputy behavior, and stale authority across company boundaries.
Read the security guideUseful evidence has explicit limits.
These resources treat agent identity as machine identity with an organizational and operational context. Authentication proves the workload. Authorization checks the action and recipient. Verification evaluates that evidence against current state.
None of those controls proves that an agent is honest, that its output is correct, or that a requested business action is safe. A receiving system needs its own application policy and execution controls after identity and authority pass.
TotemLayer
Verify the agent before your systems act.
See how Totem binds identity and current authority to one exact cross-company request.