5 replies worth sending, every morning
You know the leads are somewhere in your Buska feed, but scanning all of it before your first coffee is not how you want to spend the morning. This skill hands the triage to your AI agent: it pulls today's qualified signals, keeps the 5 conversations most worth answering, and drafts a reply for each. You read, tweak, and post them yourself.
get_signalsscore_lead1A warm feed and a connected agent
A Buska account with your signals configured, and the MCP server (buska.io/mcp) connected to your agent via OAuth (setup: buska.io/docs/mcp-server/setup). This ritual works best once your signals have run for a few days and the feed is warm.
2The agent pulls today's qualified signals
Each morning your agent calls get_signals with a since date covering the last 24 hours and a minScore threshold, so it only sees conversations Buska's engine has already scored and qualified. Every signal comes with the post, its author, an AI score and the reasoning behind it, which saves the agent from re-reading your whole feed.
3It shortlists the 5 worth joining
The agent ranks candidates by score first and freshness second: a great conversation from four days ago is usually a dead thread. When two candidates are close, it can call score_lead to break the tie on buying intent and ICP fit. For each finalist it writes one plain sentence on why this conversation matters today.
4It drafts a reply you would actually send
For each of the 5, the agent drafts a short reply grounded in what the author actually wrote: answer the question, share the useful thing, mention your product only when it genuinely fits. No template, no pitch in the first line. You stay the author; the draft is a starting point, not a script.
5Your five-reply morning
A five-item brief: platform, author, link, score, one line on why it matters, and a draft reply for each. You review, edit and post manually from your own accounts, because Buska never posts on your behalf. Most mornings that is a few minutes of judgment instead of an hour of scrolling.
The prompt to paste into your agent
Copy it into Claude, ChatGPT or any MCP client connected to Buska.
Every morning, prepare my daily reply shortlist:
1. Call get_signals with since = 24 hours ago (ISO date), minScore = 6, limit = 50.
2. Rank results by score, then by freshness. Keep the 5 best conversations, max 2 per platform.
3. If two candidates are close, use score_lead to compare buying intent and ICP fit.
4. For each of the 5, write one sentence on why this conversation is worth joining today.
5. Draft a short, human reply for each: helpful first, mention our product only if it truly fits.
6. Output a numbered list: platform, author, link, score, why it matters, draft reply.
I will review and post every reply myself. Never post anything on my behalf.Related documentation
Frequently asked questions
Can I change how many conversations it picks?
Yes. Five a day is just the prompt's default: ask for three on busy days, or ten if someone works the list full time. Keep it small enough that every reply stays personal, that is what makes the ritual pay off.
What happens on a slow day with fewer than 5 good conversations?
The agent shows you what actually cleared the bar, even if that is 2 or 3 conversations, rather than padding the list with weak candidates. You can lower minScore or widen the since window to cast a wider net. Buska qualifies real conversations; it does not guarantee a fixed lead count.
Which agents and plans does this work with?
Any MCP client: Claude, ChatGPT, Cursor and others, connected via OAuth to Buska's MCP server. You can try it during the 7-day free trial without a card; paid plans start at $49/mo (Starter). The signals themselves come from the keywords your team configures in Buska.