Scopebond Gateway — a policy-enforcement proxy for AI agents: prevent (deny out-of-policy, fail closed), prove (Ed25519-countersigned scopebond:receipts), kill switch. HTTP + MCP ingress on Hono.

scopebond-verify — the deterministic verdict library: violates(policy, receipts, claimed). The moat-bearing reference implementation.

Scopebond policy vocabulary — JSON Schema for the policy document and the scopebond:receipt envelope, plus test vectors.

Scopebond SDK — operator-side Ed25519 signing of agent action intents and a thin client for submitting them to a Scopebond gateway.

Scopebond for GitHub — check an agent pull request against your policy in your own Actions runner before it can merge. The boundary-lane connector's customer-side runner (no GitHub SDK; no Scopebond-held credential).

Scopebond framework plugins — one import so an agent checks your policy before every tool call and records a signed receipt. Cooperative (M0) in-process guard with Vercel AI SDK and LangGraph/LangChain adapters.

Scopebond hook for Claude Code and Cursor — checks each coding-agent tool call against policy before it runs (in-path, cooperative M0) and records a signed receipt, locally.

Scopebond MCP proxy — one policy for every Model Context Protocol tool call, in front of the servers an agent uses. In-path enforcement with a signed PEP-authorized receipt.