Find buying-signal posts on X and surface them in Slack

Build me an agent workflow that hunts for buying-signal posts on X (Twitter) and drops the good ones into Slack so a human can follow up. The goal is to find the right person at the right moment, not to automate the outreach itself.

Trigger: run on a cron every 30 minutes.

On each run, the agent should:

1. Search recent posts on X for a configurable list of buying-signal phrases. Default examples to start with: 'my zap broke', 'zapier is so expensive', 'looking for a zapier alternative', 'make.com is down', 'trying to build an ai agent', 'my agent keeps hallucinating', 'n8n vs zapier'. Treat this list as the main knob a user will tune. Use the X recent-search action to pull matching posts from the last 30 to 60 minutes so we only see fresh activity.

2. Filter the raw matches down to genuine moments. Skip retweets, threads from obvious bots or automation accounts, posts in languages other than English unless the user opts in, posts older than the lookback window, and posts that just mention a keyword in passing (for example, a marketing tweet that says the word 'Zapier'). Prefer posts where the author is describing a problem they are having right now, asking for recommendations, or venting about a vendor. Posts phrased in first person and in present tense are stronger signals than abstract commentary.

3. For each surviving post, fetch the author's profile (handle, display name, bio, follower count, profile link) so the human reviewer has enough context to decide whether to reach out.

4. Post each find into a Slack channel using the Slack bot. The default channel should be configurable (suggest #prospects-on-x). Each Slack message should include: the matched signal label (for example, 'Zapier complaint' or 'AI agent struggle'), the full post text, a direct link to the post, the author's handle and display name, a one-line bio snippet, follower count, and a short reason the agent thought this was a real moment (one sentence, plain English). Group multiple finds from the same run into a single Slack thread if that keeps the channel cleaner.

5. Remember which posts have already been surfaced so the same tweet never shows up twice, even if it keeps matching on later runs.

Hard rules: do not send DMs, replies, likes, or any outbound action on X. Do not draft outreach copy. The agent's only job is to find the person and hand them off in Slack. The human writes the first message.

Make the signal phrases, the lookback window, the Slack channel, and the filters (minimum follower count, language, exclude retweets) easy to edit later without rewriting the workflow.