tlon-workflow-doctor
Skill109 starsChanged 13 days agoMITView on GitHub
Use when setting up a machine for mobile agent work on this repository, when the tlon-workflow skill's prerequisite check fails, or when stim, agent-device, eas-cli, gh, or their skills are missing, outdated, or installed under an old name.
- Reads credentials
- Installs packages
In this repository5 other files it gives its agents.
- AGENTS.mdAGENTS.md
- desk/AGENTS.mdAGENTS.md
- pr-review-loopSkill
- tlon-pr-triageSkill
- tlon-workflowSkill
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 15
Run it unsandboxed, from the repository root (the path above is relative to it).
≈ “Common” is an estimate: at least 1 in 5 of the 372,429 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
Tlon workflow doctor
One script checks everything the mobile agent loop depends on, and installs what it can.
node .agents/skills/tlon-workflow-doctor/check.mjs # report
node .agents/skills/tlon-workflow-doctor/check.mjs --fix # install and re-checkRun it unsandboxed, from the repository root (the path above is relative to it). It exits 0 when every line is ok or note, and 1 while any line says fix.
Unsandboxed because gh auth status cannot reach the keyring inside a shell sandbox and reports a false not authenticated. It always inspects the source checkout, even when run from a worktree's copy: that is where apps/tlon-mobile/.env.local lives, and stim worktree warm copies it from there.
What it checks
| Line | Requirement | --fix does |
gh | 2.99.0 or newer (the --attach upload flag), authenticated | nothing; prints the install or gh auth login line |
node | the major version in .nvmrc, both for the node on PATH (which stim and eas run on) and for the one running this script; on Node 26, expo config cannot load app.config.ts, and eas reports only that simulator:availability printed no JSON | nothing; prints the nvm use && nvm alias default line |
stim | the stim package, 1.1.0 or newer, resolved first on PATH; a note when npm has a newer one | installs or upgrades stim; uninstalls stim-cli when that is what resolves |
stim ports | the ports command, which the loop's web and Cosmos steps take their ports from | upgrades stim |
stim remote | stim ios --remote, which runs the loop's devices on EAS Simulator, from stim 1.9.0 (the first release that gives each worktree's EAS session its own name) | upgrades stim |
eas | eas-cli installed, logged in, and EAS Simulator enabled for the project's account; when simulator:availability prints no JSON, the line carries eas's stderr | installs eas-cli, unless it was expo config that failed; eas login and account access need a person |
stim skill | stim skill in ~/.agents/skills or this repo | npx skills add appandflow/stim -g -y |
agent-device | installed, 0.21.13 or newer (the loop was last verified with agent-device 0.21.14 and stim 1.10.0; older pairs lost Fast Refresh and broke the session lease); a note when npm has a newer one | npm install -g agent-device |
agent-device skill | agent-device skill in ~/.agents/skills or this repo | npx skills add callstack/agent-device -g -y |
xcode | xcodebuild -version succeeds, so iOS builds on this machine; without it (Linux, or Command Line Tools only) a note that stim ios takes --eas-profile development-simulator | nothing |
ship login | DEFAULT_SHIP_LOGIN_URL + DEFAULT_SHIP_LOGIN_ACCESS_CODE, or DEFAULT_TLON_LOGIN_EMAIL + DEFAULT_TLON_LOGIN_PASSWORD, in the source checkout's apps/tlon-mobile/.env.local | nothing; they are credentials (see the tlon-workflow skill, Sign in) |
stim doctor | no cost finding in apps/tlon-mobile; one is a note with its fix, since it slows builds without blocking them | nothing; prints each finding and its fix |
--fix only installs global npm and Homebrew packages and adds skills under ~/.agents/skills. It touches nothing in this repository.
It deliberately does not run stim doctor --fix, which writes .claude/settings.local.json -- your own agent permission configuration -- and can delete generated Android .cxx directories. Both are decisions for the person running this, not repairs to apply on a tool's say-so. The stim doctor line prints what it found and the fix each finding names, as a note: a slow cache costs minutes, not the task.
When a line stays fix
The line carries the exact command. Three that need a person: gh auth login and eas login take the user's own credentials, EAS Simulator access is granted to the Expo account rather than installed, and the ship login values are credentials nobody but the user should type.
node can outlast its own fix. nvm use changes only the shell it runs in, and an agent session keeps the PATH of whatever launched it, so a session started on another Node stays on it even when nvm's default alias already matches .nvmrc. Restart the session from a shell on the pinned version.
The ship login note is worth clearing even though it is only a note. Without those two variables every reproduction that needs a signed-in app stalls on a 2FA code an unattended run cannot read. With them, sign-in is four taps and no typing.
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.

