Find leads in public conversations
Somewhere on Reddit, LinkedIn or X, someone is asking right now for exactly what you sell. Buska's engine detects those posts across 30+ platforms and scores them for buying intent. This skill turns that stream into a daily shortlist your AI agent prepares for you, with a suggested opener for each lead.
get_signalsscore_lead1What you need
A Buska account with your signals configured: the keywords, competitors and pain points your market talks about. Then connect Buska's MCP server (https://api.buska.io/mcp) to your agent via OAuth; it works in Claude, ChatGPT, Cursor and any MCP client. The setup guide is at buska.io/docs/mcp-server/setup and takes a few minutes. If you are new, the 7-day free trial needs no card.
2Pull the freshest qualified signals
Each morning your agent calls get_signals with minScore set around 7 and since set to yesterday. That returns the buying signals Buska's engine already detected and scored for your keywords, each with the AI's score and reasoning. You can narrow further with an intent filter such as ACTIVE or PAIN to keep only people actively looking or describing the problem you solve.
3Qualify each lead against your ICP
The agent then reads each signal against the ICP you described in your prompt: role, company size, market, whatever defines your buyer. Clear misses get dropped. For borderline cases it calls score_lead so Buska's scoring weighs in on buying intent and ICP fit before the lead makes the cut.
4Draft an opener for every lead that survives
For each qualified lead, the agent writes a short opener that references what the person actually said, not a generic pitch. Nothing gets posted: Buska never publishes on your behalf, and neither does this skill. You stay the sender, whether you reply directly on the platform or polish the draft in Reply Studio inside Buska.
5What lands on your desk
A ranked shortlist: platform, author, link to the post, a one-line summary, the score with its reasoning, and a suggested opener. Reviewing it takes a few minutes, and every reply you send lands in a conversation that is already about your problem space. Run it daily and the pipeline compounds.
The prompt to paste into your agent
Copy it into Claude, ChatGPT or any MCP client connected to Buska.
Every morning, run my Buska lead review:
1. Call get_signals twice: once with intent ACTIVE, once with intent PAIN (minScore 7, since yesterday as ISO date, limit 30 each). Merge and dedupe.
2. Qualify each signal against my ICP: [describe your buyer, e.g. B2B SaaS founders and heads of growth at companies under 50 people].
3. For borderline leads, call score_lead to check buying intent and ICP fit before deciding.
4. Output a shortlist sorted by score: platform, author, link, one-line summary of what they said, and the score reasoning.
5. Draft a 2-sentence opener per lead that references their exact words.
Do not post or send anything. I review and reply myself.Related documentation
Frequently asked questions
Does the agent post replies for me?
No. Buska never posts to social networks on your behalf, and this skill only drafts openers. You review each one and reply yourself, or refine it in Reply Studio inside the app. On the Scale plan, Buska's AI SDR agent can send email outreach from your own inbox, but only after you approve it.
Where do these signals come from, and does Buska see private groups?
Buska continuously monitors public conversations on 30+ platforms, including Reddit, LinkedIn, X, YouTube and Hacker News, for the signals you configured. It only reads public posts: no private groups, no DMs. Each mention is scored by AI for buying intent and ICP fit, which is what get_signals returns to your agent.
Do I need a specific plan or the API?
No API work is needed: the MCP server comes with your Buska account and connects via OAuth from Claude, ChatGPT, Cursor or any MCP client. You can test the whole flow on the 7-day free trial, no card required. The REST API and the AI SDR agent are Scale plan features, but this skill does not depend on them.