Releases: sam1siam/agentic
Release list
Agentic 0.1.0-draft.3 — Hosted testing platform
Agentic now provides a hosted HTTP/PostgreSQL recovery sandbox, read-only public URL audits, private evidence history, public MCP tools, A2A 1.0 testing tasks, and an Agent Auth provider built with the upstream Better Auth plugin. The generator imports OpenAPI and downloads a starter ZIP.
Use https://ruagentic.org/platform/ to run tests and https://ruagentic.org/connect/ to connect an agent. This release includes a standalone Node 24 CLI tarball, an adoption skill ZIP, and a reusable GitHub Action (sam1siam/agentic@v0.1.0-draft.3). Install the CLI with npm install -g ./ruagentic-0.1.0-draft.3.tgz.
Validation: 28 automated checks passed, including PostgreSQL isolation, real HTTP recovery, SDK interoperability, replay rejection and revocation. Eight production checks passed on ruagentic.org. The verification record is https://ruagentic.org/reports/platform-live.json.
The profile format remains 0.1.0-draft. All test evidence is project-authored; no independent adoption or certification is claimed. The sandbox offers autonomous synthetic testing, not human account delegation or customer transactions.
Agentic 0.1 draft 2: brand, docs, and pilot toolkit
Agentic now has an ink-and-blue identity, original logo assets, a profile generator, expanded documentation, and public independent-pilot enrollment.
The release adds Markdown page alternatives, agent documentation discovery links, a local initializer that refuses overwrites, and a seven-scenario runner for separately implemented consumers. The protocol remains 0.1.0-draft.
Validation: 17 tests, type checking, eight-route static build, and all seven reference-adapter scenarios passed. GitHub checks: https://github.com/sam1siam/agentic/actions/runs/34145035081
Website: https://ruagentic.org
Pilot enrollment: #1
Reference code and its test results are project-authored. No independent pilot, production adopter, or certification is claimed.
Agentic 0.1 ? experimental draft
Agentic 0.1 is an experimental OpenAPI action profile for tracking requests, verifying outcomes, and recovering from interrupted submissions.
Includes the versioned specification and JSON Schemas, profile validator, browser recovery lab, Node and Python consumers, a loopback HTTP/SQLite service, conformance tests, governance, and a pilot kit.
Validation: 14 behavioral/integration tests passed locally and in GitHub Actions; TypeScript checks and all five static page exports passed. The deterministic comparison covers seven scenarios and four strategies, including an existing verification workflow.
This is a prerelease proposal. Both consumers are project-authored; independent adoption and production reliability are not established. The browser lab is simulated; the local test service exercises actual dropped HTTP responses and crash recovery. See docs/CONFORMANCE.md for limits.