OpenMaple v0.2.1 launch: local Docker managed agents with one setup script #30
Replies: 2 comments
-
|
Contributor entry points added for anyone who wants to help shape the provider roadmap:
These are intentionally scoped as evaluation and adapter-boundary work, not broad production-ready claims. If you are evaluating OpenMaple for an internal platform, comments on which provider matters first are more useful than generic feature requests. |
Beta Was this translation helpful? Give feedback.
-
|
New public product tour is live on the website, YouTube, and README: https://dragonforce2010.github.io/openmaple/#tour YouTube: https://www.youtube.com/watch?v=zYhgkFomZ7M Local evaluation now has a self-contained Docker Compose path: console, API, MySQL, local dev login, local Docker runtime pools, and local Docker sandbox pools. E2B and veFaaS credentials are not required for the default local path; model keys are only needed for real model-backed loops. The tour is built from the running console and existing end-to-end launch screenshots: agent creation, cited sandbox sessions, vaults, CLI, SDK, skills, deployments, and observability. |
Beta Was this translation helpful? Give feedback.
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
-
2026-06-22 update: one-command local Docker setup
OpenMaple v0.2.1 is live: https://github.com/dragonforce2010/openmaple/releases/tag/v0.2.1
Post-release update: PR #58 has also landed on
main, with HD public proof assets for the local-Docker path: 5120x2880 real UI screenshots plus a 2560x1440 walkthrough covering setup, smoke checks, workspace settings, runtime pool drawer, sandbox pool drawer, sessions, and quickstart UI:#58
The latest
mainchecks for CI, GitHub Pages, and Cloud Deploy are green.Fastest evaluation path:
Then open:
You can also start from GitHub Codespaces: https://codespaces.new/dragonforce2010/openmaple?quickstart=1
The default local path starts separate
web,api, andmysqlservices, local dev login, local Docker runtime pools, and local Docker sandbox pools. It does not need E2B, veFaaS, or OAuth credentials; model keys are only needed for real model-backed agent loops.What is in the repo today:
Links:
OpenMaple is not an official Anthropic product. It is an open implementation of the same platform idea: split AgentRuntime from SandboxRuntime, persist session state, isolate credentials, and keep provider layers replaceable.
Feedback wanted:
Agent -> Environment -> Session -> Event Logmodel match how your team would build an internal agent platform?If this direction is useful, star the repo so I can prioritize public docs, examples, and provider adapters around real user interest.
Beta Was this translation helpful? Give feedback.
All reactions