Skip to content

docs(j5): a Mobile section in j5-live-test, and test-t3-mobile points at it - #838

Closed
Jacksondr5 wants to merge 2 commits into
j5/mainfrom
j5code/mobile-test-rig-skills
Closed

Jacksondr5 wants to merge 2 commits into
j5/mainfrom
j5code/mobile-test-rig-skills

Conversation

@Jacksondr5

Copy link
Copy Markdown
Owner

Problem

The mobile rig is on trunk: --mobile ios|android, mobile-open, stop (#825) and mobile-build (#827). An agent only uses what a skill tells it to, though. j5-live-test had nothing on mobile, and upstream's test-t3-mobile still said to run vp run dev and pair through pair-client.sh, which in J5 leaves the app empty. This is item W5 of the Mobile catch-up milestone.

What changed

  • j5-live-test gets a "Mobile" section: the five steps (setup, device_open, mobile-open, drive, teardown), a row in "Other setups", and the points that tripped testers up in live runs:
    • one iOS tester at a time, with Android as the parallel path;
    • open an Android emulator by its serial, never by its AVD name;
    • call device_open only after setup has printed the device;
    • end with agent-device close before device_close and stop, since a session left open holds the device against the next tester;
    • mobile-build when the client isn't cached;
    • what the MacBook needs, and what its "Install Failed: Authorization is required" message means;
    • the slow first device_open, and the Android and iOS input quirks.
  • test-t3-mobile (upstream's) is edited in place, keeping its structure. Each section gains a J5 pointer: set up with --mobile, launch and pair with mobile-open, and don't use pair-client.sh in J5. Teardown is the AgentDevice close, then device_close and stop. This mirrors what case 68 did for test-t3-app, and is recorded as FORK.md case 97 with a file-table row.

Live testing

Each line of the Mobile section comes from a live run.

  • The crew's tester followed earlier versions of this text cold, from fresh worktrees, on Android and iOS. Reports are under test-reports/j5code-mobile-test-rig-skills/ (aea0a411ee, 18e8929839, 5fcbc41744, 2507fa4e7a, 8cefc67d47).
  • The live-test crew's mobile testers are using this branch now.
  • Its later lines record what those runs found: the device_open timing, the agent-device close teardown, the Xcode component, and the one-tester limit.

Nothing here changes runtime behaviour.

Upstream impact

  • .agents/skills/test-t3-mobile/SKILL.md: recorded as FORK.md case 97, with a file-table row in this PR. It is an edit in place, mirroring case 68.

Checklist

  • One concern: the description has no "also"
  • Tests cover the changed behavior: not applicable, agent-direction text only
  • UI changes: none
  • Live testing: the section above says what ran live
  • Upstream-owned files: each one is recorded in FORK.md (case text and file-table row) in this PR
  • Upstream product: no change to what upstream's product does
  • Surfaces: agent skills only
  • Docs: this is the docs change; no user docs, since mobile live testing is contributor tooling

Built by Claude Opus 5.5 in Claude Code (J5 Code crew seat).

🤖 Generated with Claude Code

… at it

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@vercel

vercel Bot commented Oct 31, 2026 •

Copy link
Copy Markdown

The latest updates on your projects. Learn more about Vercel for GitHub.

Project Deployment Actions Updated
j5-code Ready Ready Preview Oct 31, 2026 10:53pm UTC

Request Review

@coderabbitai

coderabbitai Bot commented Oct 31, 2026 •

Copy link
Copy Markdown

Warning

Review limit reached

  • Run on-demand review

This review includes 3 billable files and costs up to $0.75.

  • Ask an admin to make reviews automatic

Open in CodeRabbit

Reviews can continue after your included limit without a manual trigger. An admin must approve usage-based billing.

Or wait 35 minutes for your next included review.

Check out review usage here.

View limit details

Limit details: You’ve used all 2 included reviews currently available. Your 51 included PR review attempts over the past 7 days set your current allowance at 2 reviews per hour.

Learn how review limits work.

Review configuration:

Run configuration
Commits

Reviewing files that changed from the base of the PR and between 66c3105 and 065caf7.

Files selected for processing: 3
  • .agents/skills/j5-live-test/SKILL.md
  • .agents/skills/test-t3-mobile/SKILL.md
  • FORK.md
  • Autofix · Keep fixing CodeRabbit findings and required CI, and resolving merge conflicts

@github-actions github-actions Bot added the vouch:trusted PR author is trusted by repo permissions or the VOUCHED list. label Oct 31, 2026
@github-actions github-actions Bot added the size:M 30-99 effective changed lines (test files excluded in mixed PRs). label Oct 31, 2026

@Jacksondr5 Jacksondr5 left a comment

Copy link
Copy Markdown
Owner Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Combined review from both reviewers (Claude Opus 5.5 and GPT Sol 6.1), at 0ecb6960f9. We read the skill text against the merged rig code and found it accurate:

  • the claim and cap refusals;
  • the emulator's "still booting, run again";
  • opening by the emulator-<port> serial;
  • mobile-build: commit first, one iOS build at a time;
  • the offline message;
  • the new row in the setup table.

A cold agent can follow the Mobile section. Case 97 is free on trunk, and its file-table row mirrors case 68.

One fix (P3, accuracy). FORK.md case 97 says "teardown is device_close, then its stop". The PR's own skill text now closes the AgentDevice session first (j5-live-test step 5, and test-t3-mobile's teardown: "In J5, close the AgentDevice session first"). Case 97 should give the same order: AgentDevice close, then device_close, then stop. Otherwise whoever carries it through an upstream advance gets the old order.

Optional nits:

  • Upstream line in "Open the device". In test-t3-mobile, the J5 sentence is joined onto upstream's Call device_list, then device_open … line, which rewraps it. Putting the J5 sentence on its own lines leaves that upstream line byte-identical, so upstream syncs are a little cleaner.
  • Cache miss wording. "a native dependency changed" isn't the only cause of a missing dev client. A fresh host or a pruned cache misses too. The prescribed action (mobile-build) is right either way.

This is agent-direction text, so Jackson reads it before merge.

…pstream's line stays intact

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

@Jacksondr5 Jacksondr5 left a comment

Copy link
Copy Markdown
Owner Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Combined re-review from both reviewers (Claude Opus 5.5 and GPT Sol 6.1), at 065caf7316: resolved, no remaining findings.

  • Case 97 now gives the skill's teardown order: AgentDevice close, then device_close, then stop.
  • test-t3-mobile, "Open the device". Upstream's Call device_list… line is byte-identical to trunk again; the J5 sentence is its own paragraph.
  • Cache miss. The line now also covers a host that hasn't built the dev client yet.

Ready for Jackson to read as agent-direction text.

@Jacksondr5

Copy link
Copy Markdown
Owner Author

Closed: the developer rejected this text. It encoded one person's machines and edited upstream's test-t3-mobile. #858 replaces it with a Mobile section in j5-live-test that works on any machine, now that the rig reads a machine-local config (#841).

This branch was successfully deployed

1 active deployment
Preview — 065caf73 Deployed Oct 31, 2026 by vercel[bot]
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

size:M 30-99 effective changed lines (test files excluded in mixed PRs). vouch:trusted PR author is trusted by repo permissions or the VOUCHED list.

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant