Content-blind. Sub-15ms. 13 framework integrations.
Get started →
AI agents send emails, move money, and call APIs. No framework verifies what they do. EYDII checks every action before execution — without seeing your code, prompts, data, or outputs. Policies enforced outside the agent. Independently verifiable with Ed25519.
from veritera import EydiiVerify
eydii = EydiiVerify(api_key="vt_live_...")
result = eydii.verify_sync(action="database.drop", agent_id="bot", policy="safety")
# result.verified = False
# result.reason = "Action 'database.drop' is blacklisted"| Package | Install | |
|---|---|---|
| Python | eydii-python | pip install veritera |
| TypeScript | eydii-js | npm install @veritera.ai/eydii-verify |
| LangChain | eydii-langchain | pip install langchain-eydii |
| MCP | eydii-mcp-server | Claude Code, Cowork, ChatGPT, Cursor |
13 framework integrations including OpenAI, CrewAI, LlamaIndex, n8n, Mastra, Google ADK, Agno, and Pydantic AI.
