Setting up the loom agent

The right setup order

• +1. Connect your X account (for publishing)
• +2. Train "my voice" — loom will write in your voice
• +3. Add at least 3–5 sources: 2–3 high-signal X accounts in your field + 1–2 RSS/news sources (a single source produces single-type drafts; a mixed set brings variety)
• +4. Choose your mode — "approval" (default): loom drafts, you approve; nothing is published unapproved
• +5. Set the daily draft limit and, if you want, a daily credit budget
• +6. Connect Telegram — so drafts land in your pocket with their content (optional)

Task center: assign tasks, lean back

On the Tasks page you see every standing instruction you've given loom on one screen: which accounts are watched, how many drafts each task produced in the last 7 days, how many you approved, when it last ran. You can paste a list of @accounts into the box and assign dozens of watch tasks at once.

Two modes — everything with your approval

• +approval (default) — loom watches your sources and drafts; every draft comes to your approval. Publish, edit, schedule to your best time, or reject with one click.
• +manual — loom doesn't produce on its own; it only drafts when you assign a task (task center / from inspiration)

No content is ever published without your approval — that's not a setting, it's the product's design. loom posts what you scheduled when the time comes; but you approved that content.

Safety and the brake

• +pause: with one click, pause for 1 hour or until the end of the day — scanning, generation, and the posts loom scheduled all stop; the ones YOU scheduled manually are unaffected
• +decision log: from which source, with which memory records, with which score, and with which decision each draft was generated is recorded step by step
• +designed in line with X's rules: no action that touches other accounts (replies/quotes) is ever posted automatically, daily caps are human-scale and built into the product; pods, automated DMs and follower inflation don't exist here

Breaking-story surge: loom drafts instead of notifying

When several news sources converge on the same story within a short window (e.g. a major model announcement), loom treats it as a "news burst" and starts producing immediately: quote + replies on the story's strongest fresh tweets, plus your own take on the source article — 5 drafts per story. The first hour is when visibility peaks.

• +It ignores your daily draft cap — the limit is 5 per story, so it knows where to stop. A full cap is never a good reason to miss the day's biggest story.
• +If your topic focus is set, bursts outside your focus won't bother you.
• +Drafts land in your approval queue as always — nothing is posted automatically.
• +At most 3 stories per day count as a burst — on very busy news days a 4th story gets no pack (a noise brake).
• +No notification is sent: the news value is in the drafts themselves, not in a ping.
• +You can turn this off in settings (on by default).

Engagement rotation: loom cycles the tiers

Every account you watch has an "engagement" tier (core = often · rotation = balanced · discovery = rare). Turn on "engagement rotation" in settings and loom manages those tiers itself every day, so the flow never piles onto the same handful of accounts.

• +A newly added account goes to core: 7 days of frequent engagement so it gets properly measured.
• +When that runs out it drops to rotation, then discovery — the turn passes to others.
• +At the end of a cycle loom looks at what you actually shared: if you shared, the account loops back to core; if none of its drafts were shared, the account is paused (NEVER deleted — you can re-enable it from tasks).
• +If an account was quiet and there isn't enough measurement, no decision is made — accounts are never dropped on a fake signal.
• +Core holds a limited number of accounts at a time, and only a few moves happen per run, so changes stay gradual and legible.

Follow radar: loom finds who to follow, you do the following

Turn on "follow radar" in settings and loom finds accounts posting viral content in your niche, then builds a short "worth following" list a few times a day: the account, why it was picked (how many viral posts, average engagement) and a profile link. The decision and the click are entirely yours.

Suggestions appear in the "follow suggestions" card on the right rail of your command center; marking one as followed or not interested clears it from the list. If Telegram is connected, the same list also arrives there.

• +loom NEVER auto-follows and NEVER auto-unfollows — this is a deliberate product decision.
• +Why: X's rules explicitly ban bulk following and the "unfollow whoever didn't follow back" loop (aggressive following / follow churn). That behaviour can get your account labelled or suspended.
• +X's technical ceiling is 400 follows/day (1,000 with Premium) but that is not the safe line: the violation threshold is behavioural, not numeric — high volume relative to your account size, plus low reciprocity, is what stands out.
• +No account is suggested twice, and accounts already on your watch list never appear.
• +Suggestions come from the viral pool you already scrape — no extra cost.

Auto-plug: auto-reply when a tweet takes off

Turn on auto-plug in settings and set a like threshold (e.g. 25). On the approval card, use "add auto-reply when it takes off" to write a plug for that specific tweet/thread (e.g. your product or newsletter link). When the tweet passes that threshold, loom auto-posts this text as a reply below it.

• +Only fires on tweets YOU added a plug to — not every tweet.
• +It's a reply to your own tweet, so it's ToS-safe (not spam).

Reaction GIFs

When loom produces a witty draft, it sometimes suggests a reaction GIF too (e.g. “this is fine”). On reply/quote drafts the suggestion shows on the approval card — search the term with the GIF button on 𝕏's reply screen and add it in 2 clicks. On your own tweets, the “pick a GIF” button on the approval card shows the search results in a small grid: you SEE the GIFs and PICK the one you like — your pick goes through the safety check, attaches to the card, and is used at publish. If you don't pick one, a GIF is attempted automatically at publish; if that fails, the tweet goes out without it — nothing breaks.

The same picker exists in manual generation: choose the humorous tone in /uret and the suggestion appears on the output card. On tweet and article outputs you pick from the results via “pick a GIF” (the chosen GIF goes through the safety check and lands in the media strip; publishing switches to the media path); on replies/quotes you search the term via 𝕏's GIF button.

How loom learns you

Every approval, rejection, and edit is written to memory; if you state a reason with one click when rejecting (off-topic · wrong tone · cliché...), the learning sharpens. Patterns you reject repeatedly go on the "avoid" list, your real X performance shapes generation, and every Monday morning you get a weekly letter from loom: what I learned, where I'm weak, what I suggest.