Skip to content

feat(rf-maestro): add community plugin for MaestroLibrary - #17

Open
ai94iq wants to merge 3 commits into
MarketSquare:mainfrom
ai94iq:feat/rf-maestro
Open

ai94iq wants to merge 3 commits into
MarketSquare:mainfrom
ai94iq:feat/rf-maestro

Conversation

@ai94iq

@ai94iq ai94iq commented Nov 15, 2026 •

Copy link
Copy Markdown
Contributor

What this changes

Adds rf-maestro, a community-tier plugin with one skill for robotframework-maestrolibrary: Android and iOS tests driven by Maestro, with no Appium server. I maintain both the plugin and the library.

  • plugins/rf-maestro: catalog.yaml (community, active), plugin.json, SKILL.md, 7 references, 3 example suites
  • rebased on main; no CODEOWNERS change (maintainers are in catalog.yaml)
  • follows 5.0.0: plugin version from the generator, rf-core dependency by name, no metadata.version in the skill
  • CI: "maestro": "MaestroLibrary" in SKILL_LIBRARIES of check-skill-keywords.py, and robotframework-maestrolibrary on both CI install lines
  • install.md uses robotcode doc keywords MaestroLibrary; the sentence about unsigned standalone Studio builds is removed

AI assistance

Claude Code (Claude Opus) made the edits and ran the checks. I reviewed the diff and the check output, and I have the right to submit it.

Before you open it

  • Robot Framework capabilities go through robotcode; gaps are reported upstream.
  • Generated files were regenerated with their command, not edited by hand: python scripts/build-catalog.py and python scripts/build-agent-variants.py.

For the reviewer

Maintainer per catalog.yaml: @ai94iq. The example suites were dry-run only, not run against a live device.

Local checks: check-skill-keywords.py --require-all: maestro 11 files, 0 findings, RESULT OK. pytest tests --ignore=tests/eval green on the installer, description, structure (MaestroLibrary dry runs included) and command suites; catalogue, governance, skill validation, plugin references and docs voice pass.

@manykarim

Copy link
Copy Markdown
Collaborator

Hi @ai94iq, thanks for this plugin! It's careful work. I checked the keywords and arguments it teaches against the MaestroLibrary source (Swipe with its keyword-only args, Scroll Down ... x=, Open Application ... permissions=, Run Flow with tags and **env, Expect Element, Register Keyword To Run On Failure, the import arguments) and found no mismatches. It's also great that the agent workflow uses robotcode throughout. 👍

A few steps to get it merged. Most of them come from changes on our side since you opened the PR:

Rebase on main

Since you opened this, we simplified governance: .github/CODEOWNERS now lists only the repo maintainers, and plugin maintainers are named in catalog.yaml (see GOVERNANCE.md, Review). That's why the branch now conflicts in CODEOWNERS. Could you:

  • rebase on main and drop the two rf-maestro lines from CODEOWNERS;
  • rerun python scripts/build-catalog.py and mention it in the PR, so the generated files match the current generator?

Let CI check the keywords

Right now CI doesn't look at rf-maestro: scripts/check-skill-keywords.py only checks skills listed in SKILL_LIBRARIES, and the CI jobs don't install robotframework-maestrolibrary, so the dry runs in tests/test_library_skill_structure.py skip. Adding "maestro": "MaestroLibrary" to the checker and the package to the install lines would let CI confirm what I checked by hand. If you'd rather not touch the workflow from a fork, just say so, and we'll do that part.

Two small content points

  • references/install.md:45 builds libdoc with python -m robot.libdoc. robotcode doc keywords MaestroLibrary would match the rest of the skill (RobotCode first).
  • references/flows-and-studio.md:50 points to unsigned standalone builds and how to open them. Since other people's agents read plugin text, could you drop that sentence and keep the uv add "...[studio]" route?

For the description

It's a nice fit that you maintain both the plugin and the library; please mention that in the PR, along with the note that the examples were dry-run only. We published AI_POLICY.md after you opened this PR, so you couldn't have known about it. It now applies to all open PRs, including this one. Could you add the "AI assistance" section from the current PR template to the description (one line is enough, or "none" if no tool was involved) and mention that you ran python scripts/build-catalog.py?

Looking forward to the update. Thanks again! 🎉

PS: I merged #13, #14, #15 and #16 today. I made a few small adjustments myself so they could go in right away; each PR's last comment lists them. Thanks for those! 🎉

@manykarim

Copy link
Copy Markdown
Collaborator

Hi @ai94iq, we released 5.0.0 today, and it adds two steps to the rebase:

  • No version of its own. Every plugin and the installer now share one version, which python scripts/build-catalog.py writes into each plugin.json. Please don't set "version": "1.0.0" by hand: rerunning the generator fills in the current version.
  • Name-only dependency. Dependencies are now just the plugin name, "dependencies": ["rf-core"]. The catalogue check refuses a version range such as "^1.0", because the versions always match. As in rf-web, the README can add that Copilot CLI, VS Code and Codex don't install dependencies, so users install rf-core themselves there.

bundles/ is gone too, so the generator no longer writes files there. User-visible changes go under ## Unreleased in the root CHANGELOG.md. The rest of my last comment still applies. Looking forward to the update! 🎉

Plugin maintainers live in catalog.yaml, so CODEOWNERS is unchanged.
CI now checks rf-maestro keywords and dry-runs its examples.
@ai94iq

ai94iq commented Nov 20, 2026

Copy link
Copy Markdown
Contributor Author

Thanks @manykarim, all done: rebased on main with no CODEOWNERS change, catalog regenerated, maestro added to SKILL_LIBRARIES and to the CI installs, install.md uses robotcode doc keywords, and the standalone-builds sentence is gone. It also follows 5.0.0 now (dependency by name, no metadata.version). Locally check-skill-keywords.py --require-all reports maestro with 0 findings.

Eval coverage needs a narrow task, a partial-install task and a trigger set for every shipped skill. Adds the sut-maestro fixture (Demo Login app without the Appium server) and counts rf-maestro in the partial-install test.
The security gate's credential-access rule matches the word keychain anywhere, so the Clear Keychain mentions failed it as false positives.

This branch has not been deployed

No deployments
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants