When should you use it?
- You post on X most days and react to news in a clear niche
- Scrolling for something to react to eats your morning
- You want an agent to pick candidates, not to speak for you
Not for: Having the agent draft or publish quote tweets under your name; Brand accounts that need legal or comms sign-off before every post (the timing will not work); Alerts on breaking launches every few minutes (see launch-watch-alerts-you-can-post-on)
What do you need, and what do you get?
You need
- Read access to your X account: lists, search, trending, and each post's like, repost and quote counts with its age
- Three or four X lists of 30 to 80 accounts you would genuinely react to
- A few niche search terms, your score threshold, and a ceiling on how many quotes a post may already have
- An avoid list of accounts you never want your name next to, and a log of posts you have already quoted
- Where the shortlist should land: a DM to yourself, a sheet or a Notion table
You get
- One morning message with up to five posts, each showing link, score, why it scored and one angle
- A saved scan log with every candidate's metrics and score, for the weekly review
- On request, a go or skip answer before you post, based on duplicates, account safety and whether the original is still rising
How does it work, step by step?
- Build the listsMake three or four X lists of accounts you would react to anyway. Keep each to a few dozen accounts so the pool stays relevant.
- Pick a cheap judgeEach candidate needs one yes-or-no call, so use a small, low-cost model for scoring and save the bigger model for the angles. Camp uses Nace's Drex; he says in replies that another model also works.
- Define worth quotingScore on early speed (engagement in its first half hour), fit with your niche, space for an opinion rather than applause, a low count of existing quotes, and a safe author.
- Schedule the scanRun it daily before you wake up over the last few hours of posts. Keep only scores above your threshold and send the top five, fewer if fewer qualify.
- Write, then checkPick the ones you have a view on and write each yourself. Before posting, ask the agent whether you already quoted it, whether the account is safe, and whether the post is still gaining.
The prompt
The original prompt by @JamesonCamp, word for word, with a link to where it was published. Below it is our adapted version with fill-in blanks, if you want it. Works in Grok Bot, ChatGPT agent, Claude and others. Values in lime are placeholders.
Original prompt by @JamesonCamp
Every day at [6am], open my X lists [list 1, list 2, list 3], search for [niche terms], and check trending. Pull every post from the last [3 hours] with its text, author, likes, reposts, quotes and minutes since posted. Send each to Drex: "worth a QT for [me]?" with my niche: [AI, tools, launches]. Favor early velocity, room for a real take, under [15] QTs already, and safe accounts. Keep anything over [0.8].
Exactly as published by @JamesonCamp in X Article: How You Can Get 5 Million Impressions a Month on X From Quote Tweets (Grok Bot + Drex Setup).
Our adapted version (optional)
Fill in the blanks (optional; the copied prompt uses your values)
Set up a daily read-only scan of X for posts worth quoting. Never write or publish posts for me. Run at: {run_time}. Lists: {x_lists}. Niche searches: {niche_terms}. Also check trending. Look back: {lookback_hours} hours. Score threshold: {threshold}. Max existing quotes: {quote_ceiling}. Avoid list: {avoid_accounts}. My quote log: {quote_log}. Send results to: {delivery}. 1) For each post, collect the text, the author, its like and repost totals, how many quotes it has, and its age in minutes. 2) Drop authors on my avoid list and posts already in my quote log. 3) Score each remaining post with {scoring_model} on one question: is this worth a quote from me? Weigh early speed, fit with {niche_topics}, room for an opinion, and fewer than the quote ceiling. 4) Keep scores at or above the threshold, sort them, and send me at most five as one message: link, score, why it scored, and one angle on where I could add a view. Send fewer if fewer qualify. 5) Save every candidate with its metrics and score to the scan log. 6) If I share a draft with you, check the quote log and avoid list again, then say whether the original post is still gaining. Answer go or skip with a reason.
What can go wrong?
- Scoring with a large model on thousands of posts gets expensive fast. One cheap call per post is the whole point of the design.
- A shortlist padded to five with weak picks trains you to ignore it. Send fewer on slow days.
- Quoting the same post twice, for example once from your phone and once from the queue, looks careless. Record every post you quote and check that record each time.
- Very fast posts can come from accounts you would regret amplifying. Keep an avoid list and treat it as a hard stop, not a score penalty.
- Old candidates lose most of their value. If you read the list hours late, rerun the freshness check before writing anything.
Our take
The insight in Camp's article is that the hard part of a good quote tweet is choosing the post, not writing the line. He frames it as a pipeline: a wide scan, a cheap score per post, a short list with angles, and a human who writes. We like that the agent never drafts the tweet; that keeps your voice and your judgment intact, and it is the main reason we added this one. His scan prompt is specific about the signals, and we kept it word for word below. We made three changes. First, the scoring model is swappable; the article is built around Nace's Drex, but any inexpensive model that returns a confidence works, and Camp says as much in his replies. Second, an avoid list is a hard filter, not a factor in the score, so a viral post from an account you distrust never reaches you. Third, the morning message sends fewer than five on thin days rather than padding. In our dry run on eight fictional posts (threshold 0.8, ceiling 15 quotes), the agent kept three: a 22-minute-old tool launch, a 15-minute-old open-source release and a founder thread on agent pricing. It dropped a model chart with 48 quotes as crowded, an off-niche gossip post, one from the avoid list and one already in the quote log, sent three picks with an angle each, and drafted no tweet text.
Where the time goes back: The slow part of quote-posting is finding the right post while it is still fresh. Handing that search to a scheduled agent leaves you a short read and a minute of writing per pick.
FAQ
Do I need Drex or Nace?
Will the agent write or post the quote tweets?
How is this different from launch-watch-alerts-you-can-post-on?
Will this get me a certain number of impressions?
Source and credit
The original prompt on this page is by @JamesonCamp, shown word for word with a link to where it was published. The summary, notes and adapted version are ours.
- Original
- X Article: How You Can Get 5 Million Impressions a Month on X From Quote Tweets (Grok Bot + Drex Setup)
- Creator
- James Camp (@JamesonCamp)
- Published on
- X
- Original date
- Oct 26, 2026
- License
- Not openly licensed; the original prompt is shown word for word with credit and a link to the original
- Originally for
- Grok Bot
- AI dry run
- Oct 26, 2026: passed. Dry-run tested by our AI test agent on sample input. Avoid list and quote log applied as hard filters; crowded post rejected; fewer than five sent; no drafting.
AI dry run on sample data; not run with live connected accounts. The playbooks.bot test agent filled the prompt with synthetic inputs, produced the output and checked it against the page's stated outputs and guardrails. - Verified
- Oct 26, 2026: source link resolves to the original, the credited creator published it, our excerpt or prompt matches it, and the license allows our use.
Are you the creator? Claim or correct this page.