Railguard v0.1.0 — hiring proof-of-work

Railguard v1 proof-of-work

Policy-enforced execution safety for AI-agent stablecoin payments.

What ships

• Foundry: 53 adversarial tests (dual-sig, batch injection, replay, spend caps)
• SignGate: OPA/Rego policy, EIP-712 cosign, Redis reservations, Postgres audit, chain watcher
• SDK: 12 Vitest tests with cross-language sessionId / EIP-712 / receipt vectors
• E2E: scripts/e2e-happy-path.ps1 — deploy → cosign → on-chain execute → watcher ingestion → signed receipt

Reviewer path (~10 min)

1. make test
2. Skim docs/SECURITY_REVIEW.md
3. Optional: run canonical E2E script

Scope

Base Sepolia + Anvil · USDC transfer only · ALLOW/BLOCK · no Paymaster