release-testflight
Skill1 starsChanged 39 days agoMITView on GitHub
Ship a Kata Code iOS build to TestFlight. Dispatches the Mobile TestFlight GitHub workflow, watches it, triages failures, and reports the uploaded version and build number. Falls back to building on this Mac with the signed-in Xcode account. Use when asked to cut, ship, upload, or deploy a TestFlight or iOS beta build of Kata Code.
In this repository21 other files it gives its agents.
- .cursor/rules/global.mdcCursor rule
- .cursor/rules/kata-code-browser.mdcCursor rule
- .cursor/rules/lifecycle.mdcCursor rule
- .cursor/rules/open-pstack-models.mdcCursor rule
- .cursor/rules/ponytail.mdcCursor rule
- .cursor/rules/sub-agents.mdcCursor rule
- 1passwordSkill
- astra-orchestratorSkill
- eli5Skill
- readmeSkill
- rebase-kata-upstreamSkill
- resolve-reviewsSkill
- show-meSkill
- test-t3-appSkill
- test-t3-mobileSkill
- thermo-nuclear-code-quality-reviewSkill
- thermo-nuclear-reviewSkill
- thermo-runSkill
- thermosSkill
- unslopSkill
- verify-katacodeSkill
The file
We split this skill into the rules it gives an agent
1 rule. About 0 are common, 0 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈0
Distinctive · estimate
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 65
If export stalls with an idle codesign child of xcodebuild -exportArchive, macOS is waiting on a keychain prompt for the local Apple …
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Release TestFlight
The Mobile TestFlight workflow (.github/workflows/mobile-testflight.yml) and a local run both call scripts/mobile-testflight.ts. It prebuilds the production app (com.katacode.app) with OTA off, archives with automatic signing for team ZBZKKWF95G, and uploads to App Store Connect. The build number is the UTC upload minute (YYYYMMDDHHmm). Setup and background are in docs/operations/release.md#mobile-testflight.
Bare gh in this repo targets upstream. Pass -R gannonh/kata-code on every call.
1. Check prerequisites
gh secret list -R gannonh/kata-code | grep -E '^APPLE_API_(KEY|KEY_ID|ISSUER)\b'
gh workflow view mobile-testflight.yml -R gannonh/kata-code >/dev/nullCI needs all three secrets and the workflow on main. If either is missing, CI can't run, so tell the user which piece is missing and go to the local path (step 4). Point to the one-time setup in the release doc for the secrets, since creating the App Store Connect app record and the API key needs a human.
2. Dispatch and watch
Build from main unless the user names another ref. Pass -f upload=false when they want a signed build without an upload.
run_url="$(gh workflow run mobile-testflight.yml -R gannonh/kata-code --ref main 2>&1 | grep -Eo 'https://github.com/[^ ]+/actions/runs/[0-9]+')"
run_id="${run_url##*/}"
gh run watch "$run_id" -R gannonh/kata-code --exit-status --interval 60gh workflow run prints the URL of the run it created when GitHub returns one. Watch that run, not the latest one in gh run list, which can belong to someone else's dispatch. If no URL comes back, upgrade gh rather than guessing the run.
A run takes roughly 30 to 60 minutes. Run the watch in the background and wait for it to exit instead of polling.
3. Report or triage
On success, read the result line from the log:
gh run view "$run_id" -R gannonh/kata-code --log | grep -E 'Uploaded Kata Code|Exported Kata Code'Report the version, build number, commit, and run URL. Say that App Store Connect processes the build before TestFlight lists it. Do not claim the build is installable until the user confirms it appears.
On failure, find the first error with gh run view "$run_id" -R gannonh/kata-code --log-failed, then match it:
| Symptom | Cause and action |
Missing TestFlight configuration: ... | A secret or production environment variable is unset. Report the names. |
No profiles for 'com.katacode.app...', or a certificate can't be created | The API key lacks the Admin role, or the Apple account has a pending agreement. Ask the user to check in App Store Connect. |
No suitable application records were found | The App Store Connect app record for com.katacode.app doesn't exist. Ask the user to create it. |
The train version ... is closed or must be higher than the previously approved version | App Store approval closed the current version. Ask the user which version to bump to. Don't pick one. |
The bundle version must be higher than the previously uploaded version, or a duplicate build number | Another upload started in the same UTC minute. Dispatch again; the new run gets a later build number. |
Compile or pod install errors | A real build break. Reproduce it locally with node scripts/mobile-testflight.ts --no-upload and fix it on a branch with its own Linear issue. |
4. Local fallback
Use this path when CI can't run, for example when the workflow isn't on main yet or the API key isn't set up. It needs macOS with Xcode, CocoaPods, the repo installed (vp i), and Xcode > Settings > Accounts signed in to team ZBZKKWF95G. If xcodebuild reports Unable to log in with account, the Xcode session expired. Only the user can fix that, by signing in again.
export OP_SERVICE_ACCOUNT_TOKEN=... # the Expo config loads Clerk and relay settings from 1Password
node scripts/mobile-testflight.ts # add --no-upload to stop after exportIf export stalls with an idle codesign child of xcodebuild -exportArchive, macOS is waiting on a keychain prompt for the local Apple Distribution key. Ask the user to enter their login password and choose Always Allow. CI uses a fresh keychain and never sees this prompt.
Without Clerk and relay config the script refuses to upload and names the missing variables. --no-upload still builds, for checking that the app compiles and signs.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

