agentleFS
matthewdonsemail-labmatthewdonsemail-lab/log/.claude/skills/log-instagram-new-platform/SKILL.md

log-instagram-new-platform

Skill63 starsChanged 2 days agoMITView on GitHub

Bringing a new platform up, then vendoring its client. A log play (docs/useCases/instagram/newPlatform.md): Make a platform log-cli doesn't have yet, open it in the account's [AdsPower](https://www.adspower.com/share/pD25Iu) browser, sign in by hand, keep the session, read what its own client called with rea, then find the open-source client that speaks the same calls and vendor it so the platform works on the direct-API engine. Instagram was the first one done this way.

The file

We split this skill into the rules it gives an agent

1 rule. About 0 are common, 1 is distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈1

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    The client goes out like a browser: Prefer the client's own curl_cffi transport, as X, LinkedIn and Slack do.

    line 47

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 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.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Bringing a new platform up, then vendoring its client

This is the log play instagram/newPlatform. Before anything else, call the log MCP tool getGuide({ useCase: "instagram/newPlatform" }): it returns this page and makes it the session's play, so every log tool's answer says which step you are on and a send with no play loaded is refused (rules 157, 158). Then follow the steps in order.

Draft every message from that person's own thread and matt's real replies (voice with no text, search), never from memory, and never send two that read alike (sameAsSent).

Make a platform log-cli doesn't have yet, open it in the account's AdsPower browser, sign in by hand, keep the session, read what its own client called with rea, then find the open-source client that speaks the same calls and vendor it so the platform works on the direct-API engine. Instagram was the first one done this way.

How the owner asks for it: "make a new platform via the MCP for <platform>", "open <platform> in AdsPower", "find me via the gh api the most likely to go into the vendor", "make it work"

Run it from a fresh context

Say: "go on docs/useCases/instagram/newPlatform.md for <platform>". The agent reads this page and runs the steps.

The steps

#StepToolDone when
1Make the platform, dry run first, then applynewPlatform { name, hosts, home, cookieDomain, sessionCookies }, then newPlatformLive with the same and confirmLive: trueevery file written and every syntax-tree edit landed, check:platforms passes
2Restart the MCP server so it loads the new platformthe owner restarts the Claude appadsPowerSync sees the platform's accounts
3Give the account its browser profileadsPowerSyncLive { account: "<account>", confirmLive: true }the account has an adsPowerProfileId
4Open the platform in the account's browserinspectStart { account, detach: true, url: "<home>", goal }the gate and self-check passed and the home page is open
5The owner signs in by hand in that window; read where it landedinspectDo { action: "snapshot" }the platform's own answer shows the account signed in
6Save the session, then close the browser so it is written to the accountinspectDo { action: "saveState" }, then inspectDo { action: "stop" }the account's cookies hold the platform's session cookies
7Read every call the platform's client made with reacaptureDocsdocs/platforms/<platform>/ has a doc per request, each read by rea
8Find candidate clients on GitHub, by name and by the exact operations recordedgh api search/repositories, gh api search/code for the recorded operation namesa short list, each with stars, licence and what API it speaks
9Compare how maintained each isgh api repos/<repo>/commits, releases, closed issues in the last 30 and 90 daysa table the owner picks from
10Vendor the owner's pick and wire the engine's session check to itvendor/<client> with UPSTREAM, python/<platform>Api/engine.py ping, the route's library, the platform's specthe session check passes offline tests and the gates
11Check the session through the engine<platform>SessionCheck { account }the platform's own answer: loggedIn as the account's handle

What the runs taught

  • No proxy unless the owner says so. A new platform defaults to needing a proxy in the account's home country. An account with no proxy and no home can't open until the platform is set to proxy: "none" or the account gets both.
  • The MCP server keeps the old platform list until it restarts. Sync and inspect on a platform made in the same session see none of its accounts; restart the app, never run log by hand around it.
  • Outside the account's hours is the owner's call. inspectStart refuses with outsideActiveHours; bypass with confirmBypass only when the owner says to open it now.
  • The session reaches the account only when the browser closes. saveState writes the browser's state to the run; stop saves it to the account.
  • Search by the operations, not only by name. Searching GitHub code for the recorded GraphQL operation names finds the clients that speak the same web API; a name search mostly finds mobile-API clients and bots.
  • The client goes out like a browser. Prefer the client's own curl_cffi transport, as X, LinkedIn and Slack do.

Download now (ZIP)

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
What did you use it for? e.g. Reviewing pull requests in a Rails app

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.