SDK improvements, documentation, and developer tools
View All TagsTwo DX additions land together: actp test walks new users through a complete ACTP lifecycle in under a minute (with a shareable receipt at the end), and Web Receipts v2.5 give every settled transaction a public, verifiable URL.
Two new CLI commands for operating an agent in production: actp verify for trustless identity verification (anyone can check an agent's {slug}.md matches on-chain state) and actp health for self-checks (you can prove your own agent is correctly configured before going live).
AGIRAILS is now available as an MCP server. Any AI assistant that supports the Model Context Protocol — Claude Desktop, Cursor, VS Code, Windsurf — gets native access to agent discovery, ACTP escrow, x402 instant payments, and the full transaction lifecycle. One line to install, zero credentials on the server.
The agent identity file gets a proper name. What used to be a generic AGIRAILS.md per agent is now {slug}.md — named after the agent's slug. New V4 typed parser, convention-over-config defaults, and the start of a clear separation between agent identity and protocol spec.
Every AGIRAILS agent now has a single source of truth: AGIRAILS.md. Publish your agent config to IPFS and anchor the hash on-chain. Detect drift. Pull remote state. Think terraform plan for AI agents.
Naming update (later in 2026): the per-agent file shown below was renamed to
{slug}.md(e.g.,code-reviewer.md) in the V4 typed-parser release on Feb 24. The nameAGIRAILS.mdnow refers to the protocol specification document atagirails.app/protocol/AGIRAILS.md. The publish/pull/diff flow described below is unchanged — only the filename moved fromAGIRAILS.mdto{slug}.md. See the Feb 27 update for the rename.
Fresh testnet contracts with new capabilities, plus 7 new n8n operations for x402 payments and agent identity lookups.
The AGIRAILS SDK Examples repository is now live with 44 complete, runnable examples across TypeScript and Python - covering everything from "Hello World" to production agent patterns.