agentleFS
Sign inStart free
tonydzitonydzi/second-brain-starter-kit/skills/fb-post/SKILL.md

fb-post

Skill6 starsChanged 2 months agoMITView on GitHub

Publish an already-vetted post to a personal Facebook wall through the owner's real logged-in Chrome tab (a low-ban-risk path), rate-limit-guarded and draft-first: paste, verify what is actually in the composer, confirm after reload. Triggers: "/fb-post", "publish to facebook", "post this to my wall".

1 other project uses a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

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

  • 3

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    If the owner gave a ready draft — use it.

    line 30
  2. Very specific:

    Check that the Chrome MCP is connected: mcp__Claude_in_Chrome__list_connected_browsers.

    line 37
  3. Distinctive:

    Don't republish identical text.

    line 58
  4. Very specific:

    Don't touch login/password/2FA — if the session is logged out, that is a blocker for the owner.

    line 59

≈ “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

/fb-post — publish a post to the owner's wall (safely, draft-first)

Why. The content factory and the Facebook diary already WRITE posts in the owner's voice, but the draft still ends up parked in Saved Messages / the vault. The only missing piece was the "publish" button. This skill closes it — in the safest possible way (acting inside a live logged-in Chrome tab, not with a headless bot), with a hard volume counter.

Main rules (from Deep Research #32):

  • Publishing = OUTBOUND + PUBLIC = Tier-2 → show the final text to the owner and wait for an explicit + (or let him press "Publish" himself). Never publish silently.
  • The owner's voice = the top-tier model. Take the text from a ready draft (content-factory, facebook-diary, episode) or write it with the top model. Don't invent a new voice, and don't let a cheap model write authorial text.
  • Keep the volume low: fb_guard caps it at ≤8 posts/day. Don't repeat the same text (spam flag).

0. The fuse BEFORE the browser (mandatory)

python "$USERPROFILE/.claude/scripts/fb_guard.py" check post
  • OK post (...) → you may continue.
  • BLOCKED post: ... (exit 3) → STOP. Tell the owner the daily post limit is used up, offer to schedule for tomorrow. Do not work around it.

1. The text (owner's voice, top model)

  1. If the owner gave a ready draft — use it. If he asks "make a post out of X" — write it with the top model in his voice (see fb-diary-voice: woven narrative, self-irony, ~4000 characters for a diary entry; shorter for an announcement).
  2. Show the owner the final text and ask for a +. This is the Tier-2 gate — no explicit "yes", no next step.

2. The browser (Claude-in-Chrome, live tab)

Browser work is strictly LOCAL to this machine. Don't pull the Claude window to the front. ⛔ IP gate (owner, 2026-07-16): posting/commenting on Facebook happens ONLY from the hub (a stable IP). On another machine? Do NOT post from here — send the job to the hub as text (bus / fleet log chat). Canon: ip-sensitive-actions-hub-only.

  1. Check that the Chrome MCP is connected: mcp__Claude_in_Chrome__list_connected_browsers. If not — ask the owner to open Chrome with the extension (don't fall back to Playwright).
  2. Open/select the facebook.com tab (he is already logged in — do NOT touch login/password). If not logged in — that is a blocker, tell the owner.
  3. Find the composer: mcp__Claude_in_Chrome__find with the query "What's on your mind — create post box". Click it to open the post creation dialog.
  4. Type the text into the composer field (form_input on the ref from find). Do NOT press "Publish" yourself until the Tier-2 gate from §1 is cleared.
  5. After the owner's explicit + — press the "Publish"/"Post" button (or ask him to press it, if he is nearby and wants to). ⚠️ The dialog is two-step: compose → "Next" → Post settings → "Post". Take a screenshot of the result (computer screenshot) so the owner SEES that the post went out.
  6. Footer as the 1st comment (rule §7.7, the post body carries no links): open the published post → the "Comment" field → blocks B+C+E from _imports\content-factory\_STYLE-footer.md (live links only). ⚠️ Enter sends the comment — use Shift+Enter for line breaks.
  7. Self-like (owner's rule 2026-07-05: "the first like is always the hard one"): like your own post right after the footer comment. A standard step, no separate asking.
  8. Composer pitfall: the first click can land on Stories (the feed shifts while lazy-loading) — after navigating, wait ~2s, screenshot, then click on the fresh coordinates.

3. Record it in the counter (AFTER a successful publication)

python "$USERPROFILE/.claude/scripts/fb_guard.py" record post

4. Report

One line: what was published + link/screenshot + "posts today N/8".

Kill switches (account safety)

  • Any Facebook warning (checkpoint, "you're doing this too often", an identity confirmation request) → STOP immediately, report to the owner, do NOT retry-spam (retries are exactly the road to a ban).
  • Don't republish identical text.
  • Don't touch login/password/2FA — if the session is logged out, that is a blocker for the owner.

Related

  • /fb-reply — replying to comments (Phase 1, the same guard).
  • The counter engine: ~/.claude/scripts/fb_guard.py (shared across post/reply/dm).
  • Texts: content-factory, facebook-diary, episode; the voice — fb-diary-voice.
  • Canon: Decision Memo 2026-06-28 (the Facebook skill set), chrome-autonomy-self-drive, browser-work-on-peers-not-hub.
<!--kit-footer-->

Like this skill? It is one of 100 in second-brain-starter-kit: the second brain we built for ourselves and run every day at Palo Alto AI Research Lab. Install the whole set with npx skills add tonydzi/second-brain-starter-kit. Everything is open source and free, so take what you need.

Flagships worth a look on their own: secondop-panel (a second opinion from a panel of external models), claude-memory-tidy (stop your agent's memory from rotting), telegram-mcp-kit (your own Telegram over MCP in about 15 minutes).

Author: Anton Dziatkovskii, Palo Alto AI Research Lab. Telegram @tonydzi - WhatsApp +1 341 222 9178 - X @Tony_Stef_

Engineers: want to test-drive this setup? Message me. I hand out free starter seeds to engineers who test and report back, and custom skill requests are welcome.

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