Vorim AICryptographic identity, scoped permissions, and signed audit trails for AI agents

Vorim AI is a security and identity platform for AI agents, providing each agent with a cryptographic identity, scoped permissions, and tamper-evident audit trails. It addresses the problem of agents operating with shared secrets, broad API keys, or no verifiable identity — gaps that become risky as autonomous agents take real actions in production systems. The platform is built around a five-stage agent lifecycle. On provisioning, each agent receives a unique Ed25519 keypair, with the public key and SHA-256 fingerprint becoming its permanent identity. Permissions are scoped to specific tasks rather than roles, using seven hierarchical scopes (read, write, execute, transact, communicate, delegate, elevate) that are time-bound and rate-limited by default. Every tool call passes through a permission check before execution, and both allowed and denied actions are logged. Audit trails are signed and hash-chained, producing tamper-evident bundles that can be exported for compliance reviews such as SOC 2 or the EU AI Act. When a task ends, credentials are automatically revoked to avoid lingering service accounts. Vorim markets a sub-5ms permission check latency (p99) achieved through Redis-cached verification. It targets platform leads, staff engineers, security architects, and GRC/compliance teams at organizations deploying AI agents. Vorim positions itself as integrating with common agent frameworks and tools — including LangChain, OpenAI, Claude, CrewAI, LlamaIndex, MCP, A2A, AutoGen, and automation platforms like Zapier, Make, and n8n — and emphasizes a lightweight, three-lines-of-code integration via an open protocol. Additional capabilities include trust scoring (live 0-100 scores with ranked reason codes and signed attestations), credential delegation that lets agents access OAuth services like Google, GitHub, and Slack through scoped grants without exposing refresh tokens, and runtime controls that route high-risk actions to a human approval queue. As a relatively new and specialized product, Vorim's main trade-off is that it adds an infrastructure layer to an agent stack and requires adopting its protocol and lifecycle model. Buyers should validate the maturity of its integrations and compliance claims against their own requirements.