AgentContract v0.1.1 — First Production-Ready Protocol Release

AgentContract v0.1.1 - Production Ready

This release marks the first stable, production-ready version of AgentContract, a security-first protocol that enables websites to publish cryptographically verifiable, agent-readable truth for AI systems.

AgentContract introduces a deterministic interface layer designed for the emerging agentic web. It allows AI agents to securely discover trusted data, verify integrity, and interact with websites without relying on scraping or risking hallucinated outputs.

Highlights

• Ed25519 cryptographic verification for trusted agent communication
• Zero-shot tool generation for OpenAI-compatible function schemas
• Fail-closed security architecture that prevents unsafe fallbacks
• Canonical agent.json specification for deterministic site truth
• Monorepo ecosystem with modular SDK packages

Included Packages

• @anmollcodes/web — Define and serve agent.json
• @anmollcodes/agent — Discover and verify agent truth
• @anmollcodes/core — Shared schemas and cryptographic primitives

Developer Tools

• CLI Inspector for AgentContract compliance validation
• Schema validation and signature verification
• Performance-optimized discovery workflow

Improvements in This Release

• Hardened security model
• Professional documentation structure
• CI/CD integration
• Protocol architecture refinement
• Package ecosystem stabilization

Vision
AgentContract aims to become the canonical trust layer between websites and AI agents, replacing unreliable scraping mechanisms with cryptographically verified, intent-based interfaces.

Documentation
Refer to the README and PROTOCOL.md for implementation details and technical specifications.