agentleFS
Sign inStart free
DezXBTDezXBT/AgentX/AGENTS.md

AgentX

AGENTS.md47 starsChanged 5 months agoMITView on GitHub

This file tells an autonomous agent how to drive X/Twitter through agentx. - Drive everything through the agentx binary (or import pkg/agent in Go). - Every command prints one JSON object to stdout and nothing else: - success: {"ok":true,"schema_version":"1","data":...,"pagination":{"nextCursor":"..."}} - failure: {"ok":false,"schema_version":"1","error":{"code":"...","message":"..."}} - process exit code is 0 on success, 1 on a handled error. - Parse the envelope; never scrape human text. data shape depends on the command (a tweet array for timelines, a user object for user, etc.).

  • Where files are headingNew AGENTS.md files have more rules about keeping secrets out of the code, and fewer about code style.

    Of the AGENTS.md files we read, the 31,774 first seen in the last six months, against the older ones: the kind of rule that gained most share, and the kind that lost most. About files like this in general, not this one, and an estimate, since a kind is a pattern in the words.

  • In this repositoryOne other file it gives its agents.

The file

We split this AGENTS.md into the rules it gives an agent

3 rules. About 0 are common, 1 is distinctive, and 2 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.

  • 2

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Every command prints one JSON object to stdout and nothing else:

    line 8
  2. Very specific:

    Prefer supplying the full browser cookie string via --cookie: it keeps the session looking like a real browser, which makes the cookies …

    line 38
  3. Very specific:

    Keep one stable User-Agent per account (set --user-agent at add time) — do not rotate it between requests.

    line 64

≈ “Common” is an estimate: at least 1 in 5 of the 40,246 AGENTS.md 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
AGENTS.md

agentx — agent integration guide

This file tells an autonomous agent how to drive X/Twitter through agentx.

Contract

  • Drive everything through the agentx binary (or import pkg/agent in Go).
  • Every command prints one JSON object to stdout and nothing else:
    • success: {"ok":true,"schema_version":"1","data":...,"pagination":{"nextCursor":"..."}}
    • failure: {"ok":false,"schema_version":"1","error":{"code":"...","message":"..."}}
    • process exit code is 0 on success, 1 on a handled error.
  • Parse the envelope; never scrape human text. data shape depends on the command (a tweet array for timelines, a user object for user, etc.).

Decision guide

GoalCommand
Read my home feedagentx feed -n N (add --latest for chronological)
Searchagentx search --type Top|Latest "<query>" -n N
Look up a profileagentx user <handle>
Read someone's tweetsagentx posts <handle> -n N
Read one tweet (no login needed)agentx tweet <id|url>
Read a tweet + repliesagentx thread <id|url>
Publishagentx post "<text>"
Reply / quoteagentx reply <id> "<text>" / agentx quote <id> "<text>"
Engageagentx like|unlike|retweet|unretweet|bookmark|delete <id>

<id|url> accepts a numeric id or any https://x.com/<user>/status/<id> URL.

Multi-account

  • Register once: agentx account add --name <n> --auth-token <t> --ct0 <c> [--cookie "<full cookie>"].
  • Choose per call with -a <name> or set AGENTX_ACCOUNT.
  • The first account becomes the default; change it with agentx account default <name>.
  • Inspect with agentx account list (returns names + which is default).

Prefer supplying the full browser cookie string via --cookie: it keeps the session looking like a real browser, which makes the cookies last much longer.

Pagination

Timeline-style results include pagination.nextCursor. To page, re-issue the same command with --cursor <nextCursor>. Stop when nextCursor is empty or stops changing.

Handling errors

Switch on error.code:

  • not_authenticated → cookies missing/expired; ask the user to re-add the account, then retry.
  • rate_limited → back off (seconds→minutes) and retry; the client already retries transient 429s a few times.
  • query_id_error / repeated not_found on a known-good id → x.com rotated a GraphQL query id; report it so the operator can refresh ids.
  • invalid_input → fix arguments; do not retry blindly.
  • network_error → transient; retry with backoff.

Notes for safe automation

  • Writes (post, like, ...) change real account state. Confirm intent before bulk actions.
  • Keep one stable User-Agent per account (set --user-agent at add time) — do not rotate it between requests.

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.

Check your setup, free