Add Claude Code tool permissions and feedback loop instructions

[mokagio]

Summary

Adds rake, xcodebuild, xcrun simctl, bundle exec, and swift package to the Claude Code team-level allow list so agents can lint, build, test, and resolve packages without manual approval.

Context: https://linear.app/a8c/issue/AINFRA-1965/add-first-round-of-permissions-allow-lists-to-apple-projects

Test Plan

• CI passes

[wpmobilebot]

📲 You can test the changes from this Pull Request in Jetpack by scanning the QR code below to install the corresponding build.

	App Name	Jetpack
	Configuration	Release-Alpha
	Build Number	30872
	Version	PR #25237
	Bundle ID	com.jetpack.alpha
	Commit	69cefa8
	Installation URL	4jh2ra0mg42fg

Automatticians: You can use our internal self-serve MC tool to give yourself access to those builds if needed.

[wpmobilebot]

📲 You can test the changes from this Pull Request in WordPress by scanning the QR code below to install the corresponding build.

	App Name	WordPress
	Configuration	Release-Alpha
	Build Number	30872
	Version	PR #25237
	Bundle ID	org.wordpress.alpha
	Commit	69cefa8
	Installation URL	5gvapgh1fm8b8

Automatticians: You can use our internal self-serve MC tool to give yourself access to those builds if needed.

[wpmobilebot]

🤖 Build Failure Analysis

This build has failures. Claude has analyzed them - check the build annotations for details.

[sonarqubecloud]

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud

[crazytonyli]

This essentially grants permission to run any Ruby script, which we probably don't want.

[mokagio]

See my thoughts in https://github.com/Automattic/pocket-casts-ios/pull/3934/changes#r2791005452.

I see your point though, maybe there's a middle ground between this and being prompted constantly?

What about bundle exec fastlane *?

[AliSoftware]

What about bundle exec fastlane *?

Note that even bundle exec fastlane allows to run arbitrary fastlane action—including the ones not exposed as a lane in our Fastfile—via bundle exec fastlane run …

(For example there may even be a way for the agent to run bundle exec fastlane run sh … to run arbitrary shell commands)

So even that might be too permissive?

[crazytonyli]

maybe there's a middle ground between this and being prompted constantly?

I don't see any issue in being prompted constantly, actually. You get many choices locally: allow it once, allow it within the session, allow it on your computer (if you want to live dangerously 😄).

[kean]

Is fastlane needed for the feedback loop?
If it's for background agents, can they have a separate set of permissions appropriate for the machine they are running on? For my personal machine, I'd like it to err on the safer side.

[mokagio]

Just a followup to thank you all for the input.

It's clear to me that I went too wide with the permissions...

In thinking about it, I begun considering adding a fine grained list of permissions (bundle exec fastlane test, bundle exec rake lint, etc.) however, maintaining such a list could quickly become tedious.

You might have noticed I closed the PR already. My current thinking (for what is worth, given I'm just learning as I go) is along the lines of @kean: it might be simpler to manage permissions on the background agent end.

[mokagio]

Closing in favor of a smaller PR that only adds CLAUDE.md instructions, decoupled from the .claude/settings.json permission changes.