agentleFS
matthewdonsemail-labmatthewdonsemail-lab/log/.claude/skills/log-twitter-notifications/SKILL.md

log-twitter-notifications

Skill63 starsChanged yesterdayMITView on GitHub

Seeing who engaged with your posts. A log play (docs/useCases/twitter/notifications.md): After a post goes out, people like it, repost it, reply, quote it and follow you. This page pulls all of that from the account's X notifications, new and back through the history, keeps it on this machine, and gives the agent a list of who did what on which post. It sends nothing.

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
  1. Very specific:

    The lists need X's live operation: If the answer is engagersOperationUnknown, X was not called: read the account's X web client once (log …

    line 45

≈ “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.

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

Seeing who engaged with your posts

This is the log play twitter/notifications. Before anything else, call the log MCP tool getGuide({ useCase: "twitter/notifications" }): 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).

After a post goes out, people like it, repost it, reply, quote it and follow you. This page pulls all of that from the account's X notifications, new and back through the history, keeps it on this machine, and gives the agent a list of who did what on which post. It sends nothing.

How the owner asks for it: "check my x notifications", "who liked my tweet", "who engaged with my post since <date>", "pull my twitter notifications".

Run it from a fresh context

Say: "go on docs/useCases/twitter/notifications.md for <account>", or "who liked <post url>", or "who engaged since <date>". The agent then reads this page and README.md and runs the steps.

The steps

#StepToolDone when
0Session checktwitterSessionCheckloggedIn
1Read what is newtwitterNotifications { account }notificationSync says caughtUp, or a page cap
2The first time only, page back through the historytwitterNotifications { account, full: true }historyDone, or run it again later for the rest
3Unroll a post: read its whole likers and reposters lists, past the few a notification namestwitterNotifications { account, post } for one post, or { account, full: true, since } for every post with an "and N others" notificationunroll.state is done and the post's likes and reposts say readToTheEnd; stopped means run it again later for the rest
4Ask what you need from what is kept, calling nothingtwitterNotifications { account, stored: true, type, since, post }the rows, counts by kind, engagerPosts for a post and engagers ranked by how much each person did
5Act on who engaged through the page for that jobengagerDms.md, engagementReplies.md, networkMaxing.mdthat page's own done

Each row is one person doing one thing: kind (like, repost, reply, quote, follow, mention, other), handle and userId, followsYou, which of the account's posts (postId, postUrl, postText), when (at), and for a reply or quote their own tweet (tweetUrl, text). type is likes, reposts, replies, quotes, follows or all; post takes a post id or url. source says where the row came from: notification, or engagements for a person only the post's likers or reposters list names (their at is null, because the list gives no time, so since leaves them out).

For each post unrolled, and for post from what is kept, the answer says how many likers and reposters came from notifications (fromNotifications), how many unrolling added (addedByUnroll), the total, and whether the list was readToTheEnd, stoppedBeforeTheEnd or notUnrolled. In text that is one line a post: post <id>: likers 154 from notifications + 87 added by unrolling = 241 (read to the end); reposters ....

What to know

  • One read, many pages, one session. The read goes through the account's one X session, its engine host, never a session of its own (rule 105), one call on the account at a time, so the queue's sends wait for it and it waits for them. Pages are 40 notifications, 4 to 11 s apart, each through the account's governor. A 429 or a governor stop ends the read and keeps what came; nothing reads again until the owner says so (rule 94). Another run sending on the same account shares the same governor.
  • X decides how far back. --full goes on from where the last one stopped, 25 pages a run by default (pages), until X gives no more.
  • Aggregated notifications are opened up, then unrolled. "@a and 12 others liked your post" is kept as one row per person X names, and X names only some of them. The unroll reads the post's whole likers list and reposters list and adds everyone the notification left out, under the same key, so a person in both is one row. The account's own actions are never kept.
  • One state machine runs the read. The feed, then which posts, then likers page by page, then reposters page by page, five pages a call, at most 40 pages a run. A 429, a governor stop or the page cap ends it stopped, said plainly, with every row and each list's place kept; the next run goes on from there. It never retries by itself (rule 94).
  • The lists need X's live operation. If the answer is engagersOperationUnknown, X was not called: read the account's X web client once (log twitter session webClient), then read again.
  • Kept, not synced anywhere. The rows live in ~/.logCli/twitter/<account>/notifications.json, never in Twenty.

What X shows

X shows the likes on the account's own posts only, never on someone else's. Its notifications name a few of the people behind "and N others"; its post engagements screen shows the author every liker and reposter, and that screen's lists are what the unroll reads. A list gives people, not times, so a person only the list names has no at. The count X shows on a post can still sit above the total kept: X counts the account's own like, and people whose accounts are gone or hidden are not in the list.

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.