A developer has released two open-source protocols designed to bring behavioral contracts and runtime integrity monitoring to AI agents working with Anthropic, OpenAI, and Gemini. The Agent Alignment Protocol (AAP) and Agent Integrity Protocol (AIP) address a critical gap in agent coordination by establishing standards for what agents can do, have done, and are thinking about doing.

Current agent protocol stacks handle capabilities and payments but lack alignment mechanisms. AAP provides structured declarations of agent values, autonomy boundaries, and escalation triggers through Alignment Cards, while AIP monitors reasoning traces to detect when agents consider forbidden actions. The system catches boundary violations in real-time—like an agent contemplating payment API access when it's explicitly forbidden—before the action occurs, not after during post-mortem reviews.

Built as Apache-licensed SDKs for npm and PyPI, the protocols include a free gateway proxy (smoltbot) that adds integrity checking with zero code changes. They integrate with OpenTelemetry for observability, supporting Grafana, Datadog, and other backends. The developer emphasizes these are transparency protocols, not trust guarantees—they make behavior observable but don't ensure agents are trustworthy. A live demo shows four agents coordinating on a production incident, demonstrating value compatibility checks before collaboration begins.