No action executes without authorization.

Runtime governance for AI agents: pre-execution authorization, fail-closed enforcement, and a signed receipt for every decision. The control plane is live and the proof is public.

npm install @11ai/execution-governance

Gate any MCP server's tool calls, no code changes:

npx @11ai/mcp-gate --help

One unauthenticated request returns a real, signed governance decision with its full EA-11 evidence chain:

curl https://control.11aiblockchain.com/v1/public/evidence

Then check the signature on your own machine, no API key and no account:

git clone https://github.com/11-11AI/verify-11ai-proof
cd verify-11ai-proof
pip install -r requirements.txt
python verify.py

That verifies the Ed25519 signature with no extra dependencies. The envelope also carries ML-DSA-87 and SLH-DSA post-quantum signatures, which are not checked by default: the script reports them as skipped rather than claiming a pass, because the envelope's own "valid" field is the server's claim and not evidence. To check all three, pip install liboqs-python and re-run.

This organisation publishes the SDK, the MCP proxy, the receipt format and the doctrine. The policy core and the hosted control plane stay closed.

Cryptography, named per chain: EA-11 evidence is SHA-512 · RFC-EG-0010 lineage is SHA3-512 and BLAKE2b-512, dual and independently verified · SDK receipts are SHA3-512, Ed25519 signed · post-quantum envelope is ML-DSA-87 (FIPS 204) and SLH-DSA-SHA2-128f (FIPS 205). Fail-closed by default.

© 2026 11 AI Blockchain Developments LLC · Apache-2.0 for released code · Patent pending