FYI working on a GitLab Issue to MR automation template that is modeled after the GitHub one. I'm expecting a prospect to ask about this soon. https://github.com/OpenHands/extensions/pull/525 Still refining it but will let folks know when ready for review.

Feedback from a prospect: It would be great to have our current Automation templates to also support GitLab out of the box

I'd love to see us add Apps as a new folder/catalog for the extensions repo github.com/OpenHands/extensions/issues/530 . Not sure if there needs to be anything established besides a folder to make PRs against so just making an issue for now

As discussed yesterday: refactor: name automation templates after the action, not the agent

github.com/OpenHands/extensions/pull/534 - add templates for custom automation creation

I'm trying to implement some apps using the new sidecar architecture, and having some issues Background: • I host my canvas at

<http://canvas.rbren.io|canvas.rbren.io>

• Apps currently serve their SPAs in an iframe, served on a different port • Apps currently serve their backends also on a different port The problem is that agent-server credentials are stored in localStorage, so the iframe window can't access them. I also disallow traffic to arbitrary ports on

<http://canvas.rbren.io|canvas.rbren.io>

, so everything has to be proxied through OpenHands, or a special nginx route has to be set up for each app. Here's my suggestion: • When an app is installed, the backend sets up a proxy for

/apps/$app-name

, pointing to

localhost:$app-port

• This serves both frontend and backend • OpenHands decides the port (so we can avoid conflicts, e.g. two different apps trying to claim 3000). App must respect some env var like $OPENHANDS_PORT during setup • If the app wants to set up both a frontend and a backend, it can start a proxy server

On one of the above, 17341, another pass with DeepSeek-v4-Pro confirms Astra: https://openhands-ai.slack.com/archives/C0BLWB8PQ3B/p1789448533870169?thread_ts=1789448115.576689&cid=C0BLWB8PQ3B

though it's a bit messy since the bot posted multiple reviews

Hey 👋 I would to do a release of the extensions in a couple of hours. If you have a PR you would like to merge: please post the PR link in 🧵

Extensions v0.21.0 is out.

Draft PR of a skill to help build apps: github.com/OpenHands/extensions/pull/593 I'm not sure what normal testing looks like for this repo so if anyone sees anything out of place LMK

👋 OpenHands/extensions#606 needs a human review: fix(repo-monitor): honor AUTOMATION_MODEL profile by @onatozmenn This contributor's PR needs a human review, so OpenHands skipped the automatic AI review.

👋 OpenHands/extensions#607 needs a human review: fix(iterate): ship real vendor manifest dirs for Codex install by @onatozmenn This contributor's PR needs a human review, so OpenHands skipped the automatic AI review.

👋 OpenHands/extensions#608 needs a human review: fix(add-skill): resolve script path and install scope in docs by @onatozmenn This contributor's PR needs a human review, so OpenHands skipped the automatic AI review.

Improvement to the issue triage automation: github.com/OpenHands/extensions/pull/614

Do we want a tiny skill to teach the agent about Jev: use-jev ?

Some of our knowledge skills, like github, linear, gitlab, bitbucket, etc. are very specific in how OpenHands should access those downstream integrations. However, with the rise/adoption of OAuth MCP servers in Canvas, should we be adopting a more generic approach that instructs the agent to look for the Secrets (e.g., GITHUB_TOKEN) or available MCP Server connection?

Hey @hundreds-twilight-32833 and @wooden-umbrella-84098, any objection to moving OpenHands/canvas-apps to the apps/ directory of OpenHands/extensions?

@average-processor-75195 Would you reviewgithub.com/OpenHands/extensions/pull/701 related to multi-image support. I have tested against a replicated install. Would you test against your helm environment?

PR to add agent canvas design guidance when building Apps. Some of the UI getting generated for apps is just too out of place. github.com/OpenHands/extensions/pull/706

The reviewer bot approved, but also wants you to review @limited-student-10747 🙏 github.com/OpenHands/extensions/pull/602

Small fix for a problem with a ready-to-run automation and KV-cache fix(review): store stale-CI state as the raw KV value The fix is currently running in the OSS-VM and it is working fine

@limited-student-10747 @hundreds-twilight-32833 Another iteration on the software-factory skills after the changes in main of yesterday. (under 2 min 😉 ) I think it should be ready as a first version. Here the PR: github.com/OpenHands/extensions/pull/775 PS: I revoked the token that I used in the video 😉

I forgot to enable the software-factory skill by default 😞 feat(skills): enable software-factory by default

I'd like to propose that all or main contributors to OpenHands give a try to this skill in PR descriptions: humanlayer/show-me It may make PR summaries more readable or more easily parseable by humans. We could see if it works, e.g. over a few weeks? How would it be best to include it for our agents, WDYT? • tell the link in AGENTS.md to use it / in review to expect it • copy it in .agents/skills and tell to use it • something else?

Fix for CI in extensions: fix(vulnerability-remediation): escape find {} in remediation prompt

When I was doing a poc of an Events view for Automations I found that the extensions repo has the route types. Wondering if this still makes sense? github.com/OpenHands/extensions/pull/719/changes#…

Waiting on the triage for the issue on this one but really excited to have an infinite canvas app github.com/OpenHands/extensions/pull/861 . I used excalidraw and added multiple board support 😍

Hey @quick-window-91705 it seems that in the transition to plugin-based automations may have messed up our standalone skills for things like issue triage: github.com/OpenHands/extensions/…/github-issue-triage Maybe we could try to revert so the skills are also usable standalone and the plugin symlinks to the skill?

This looks like good material for a skill. I can think of an immediate use case: • the autogenerated titles Also interesting to read! LLM-generated text is so tiring, and it articulates why Five Principles for Generating Readable AI Text in Bulk