Open source demos showing how Tenuo protects AI agents with cryptographic capability tokens.
| Demo | What it shows |
|---|---|
| Invoice Processing | A multi-agent pipeline processes invoices. A prompt injection redirects a payment to the wrong bank account. Traditional auth (OAuth, RBAC, policy engines) all approve. Tenuo blocks it. |
| Skyvern Prompt Injection | An AI shopping agent browses an e-commerce store. A hidden prompt injection tricks it into buying a terrible product. Tenuo's warrant constraints block the bad purchase — the LLM is compromised but the outcome isn't. |
| Temporal Incident Response | An AI on-call agent triages a production incident as a Temporal workflow. Investigation runs autonomously; the production restart is gated by a cryptographically signed human approval; the receipt chain is offline-verifiable end-to-end. |
AI agents need access to production systems — databases, APIs, payment processors. Traditional authorization (OAuth scopes, RBAC roles, policy engines) answers "is this agent allowed to do this?" The answer is yes. The problem is that's the wrong question.
Tenuo issues cryptographic capability tokens that answer a different question: "what should this agent be doing right now, with these specific arguments, for how long?" The token carries per-argument constraints derived from the task context and is enforced in microseconds with no network round-trip.
- Open source core: github.com/tenuo-ai/tenuo
- Integrations: LangGraph, LangChain, CrewAI, OpenAI, MCP, Google ADK, Temporal
- Cloud dashboard: cloud.tenuo.ai
- Early access: tenuo.ai/early-access.html
MIT