Action State Group Inc. is launching open-source tools aimed at a trust problem that arises as AI agents buy, sell and delegate work across company boundaries. EvidenceBook gives each participant its own signed transaction history, so a buyer records what it requested and paid while a provider records what it accepted and reported delivering. Comparing selected records can show agreement, conflicts or gaps without a central authority holding everyone's data.
Countersign checks specified cryptographic and structural properties of an evidence bundle, signs its findings and records them in its own transparency log. It does not certify that claimed outcomes occurred or that the evidence is complete. Terms Packs evaluate whether recorded work met defined requirements, and an initial pack covers agent-related obligations under the European Union's AI Act.
EvidenceBook is available under the Apache 2.0 license. CEO Steven Mih said Countersign is currently free and that the company will start charging once a relying party accepts the Countersigns. Early integrations include Mesh-LLM and Project Nanda, and the specifications remain individual Internet-Drafts rather than an RFC.
Why it matters if you build agents
Developers building agents that transact with outside parties can keep their own signed records of requests, payments and delivery, rather than relying only on a counterparty's logs.
Verification confirms that reported events occurred, but it does not confirm that an agreement was satisfied or that every action was captured, so teams should not treat it as a compliance certificate.
Teams using Terms Packs should customize the rules for their own requirements, since the packs do not certify legal compliance and human review remains part of the process.
Get stories like this every weekday
One short email with the day's agent tooling news.