Drop-in integrations for the agent ecosystem. Each one routes through Inherence's hosted policy gate or verifies a 128-byte receipt at the edge — your auditor, not ours, holds the proof.
Local MCP proxy. Gates every Claude Desktop / Cursor / Continue tool call against the six-rule catastrophe shield. Shipped on PyPI.
Full SDK with witnessed / attested / proven modes. OpenAI + Anthropic wrappers, MCP server, Stripe Issuing adapter, on-chain Solidity verifier.
Static linter for Vercel Zero programs. Per-program scope sidecar; one command (scope --infer) produces a tight auditor-ready manifest. Build-time enforcement.
Vercel Edge Middleware. Verifies 128-byte Groth16 receipts at the edge — sub-1ms verify, no backend round-trip. Fits trivially in the 50ms edge CPU budget.
Drop-in policy gating for Mastra agents. Wraps your tool's execute function; throws InherencePolicyDenial on out-of-policy calls. Zero deps.
Cryptographic audit trail on every agent-driven Stripe Issuing authorization. Stripe spending_controls map directly to AP2 policy parameters. Compiles and runs end-to-end.
Power-symmetric codesign — Inherence supplies the policy compiler + proof generation; the partner supplies the SDK distribution + identity / framework vocabulary. We ship the unilateral reference adapter; the canonical version emerges from the joint conversation.
WorkOS-authenticated identity → Inherence-gated agent action. Reference adapter shipped; codesign for canonical version in progress.
Auth0 Marketplace integration. Pending submission.
Remote MCP Connector for claude.ai web sessions. Application package drafted; pending Anthropic-side onboarding.
RFC draft for an MCP spec extension. Community-contribution status; working-group discussion ahead.
Every tool ships with the same architectural invariant: your auditor verifies the proof, not us.
Start a pilot → Talk to us