Lead Monitor Agent
获客线索监控 Agent
Turn "who is asking about your product right now" into a daily list of qualified leads — monitored 24/7, scored by AI, replied to in one click.
它能做什么
Lead Monitor is a locally runnable agent that watches keywords across the communities your buyers actually use — Reddit, X (Twitter), V2EX, Zhihu, Xiaohongshu, Weibo, Juejin and IndieHackers — and turns matching posts into a pipeline of scored leads. Every raw post passes through an LLM intent filter that separates "genuinely shopping for a solution" from "just asking a question", then drafts a reply you approve before it is sent. You only supply your own LLM API key, keyword list and platform credentials; everything runs from a single config.yaml behind a local dashboard. It is sold as working software — docker run or npm i, then a real crawl inside 15 minutes — not as a prompt document.
功能
- ✓ Monitor unlimited keywords across 8+ platforms (Reddit, X/Twitter, V2EX, Zhihu, Xiaohongshu, Weibo, Juejin, IndieHackers) on a 15-minute polling cycle, 7×24
- ✓ LLM intent scoring from 0–100: only posts above your threshold (default 70) reach the lead board, cutting noise by roughly 80%
- ✓ One-click AI reply drafting in your brand voice, fully editable before send, with an optional human-approval gate per platform
- ✓ Local web dashboard (localhost) showing raw post JSON, intent score and reply draft side by side for every matched thread
- ✓ Single-file YAML configuration — keywords, platforms, LLM model and reply strategy all live in config.yaml
- ✓ Editable prompts/ folder with separate intent-judgement and reply-generation templates so you can tune the agent's behavior
- ✓ Docker image plus run.sh / run.bat one-click launchers — no cloud dependency, your crawl data never leaves your machine
- ✓ demo/ folder with real sample threads walking post → intent score → reply draft end to end
- ✓ Thread-memory store in SQLite: every thread already seen or replied to is recorded, so the same post is never double-alerted and never re-answered days later
- ✓ Webhook and CSV export of the scored lead board to HubSpot, Pipedrive, Notion or a plain file, so warm leads enter your existing pipeline without copy-paste
- ✓ Per-keyword and per-platform weights — you can raise the threshold on noisy X mentions while keeping Reddit more permissive, all from config.yaml
- ✓ Weekly lead digest email: matches found, pass-rate through the intent filter and your reply activity per platform, so you can see the funnel improving over time
适用场景
- ▸ An indie SaaS founder monitors "how do I get users" plus their category keywords on IndieHackers and Reddit, and answers 5–8 high-intent posts a day that previously went unnoticed
- ▸ A B2B service agency tracks named competitors and "I need a [service] provider" phrases across X and Zhihu to jump into buying conversations within the hour
- ▸ A developer-tools startup watches Juejin and V2EX for pain-point threads matching their product and routes scored leads to the founder's Telegram before competitors reply
- ▸ A solo consultant seeds 10 keywords around their niche and lets the agent surface the one or two conversations a week that actually convert to paid calls
- ▸ A vertical B2B firm (for example a bookkeeper who serves dental clinics) monitors "looking for a bookkeeper for my clinic"-style phrases on Reddit and Zhihu and books 2–3 discovery calls a week off replies written from the dashboard
- ▸ A product team runs a competitor-intel watch: threads where users complain about a rival's pricing, missing feature or bad migration get flagged and routed to product and sales the same day the complaint is posted
交付内容
- 📦 config.yaml — single-file configuration (keywords, platforms, LLM key, reply strategy)
- 📦 run.sh / run.bat one-click launchers and Docker image
- 📦 README.md — 15-minute illustrated setup guide
- 📦 prompts/ — editable intent-judgement and reply-generation templates
- 📦 demo/ — sample posts with score and reply-draft walkthrough
- 📦 1 year of updates
FAQ
Which platforms can I monitor, and do I need paid API access for each?
Out of the box it monitors Reddit, X/Twitter, V2EX, Zhihu, Xiaohongshu, Weibo, Juejin and IndieHackers. Reddit and IndieHackers work with free or official endpoints; X and the Chinese platforms need a platform API key that you add to config.yaml — the README links the official place to get each one.
Do I need my own LLM API key?
Yes — the agent calls your own OpenAI-compatible, Anthropic or DeepSeek key for intent scoring and reply drafting. That keeps per-post cost in the fraction of a cent, gives you full data privacy, and avoids any per-seat subscription from us.
Is this an automatic reply bot that posts without me?
No. By default the agent drafts replies and waits for your approval in the dashboard. You can enable auto-send per platform if you really want, but the point is that you keep control of what gets posted under your name.
How long until it is actually finding leads?
Most buyers run a real crawl within 15 minutes of cloning: fill in config.yaml, run run.sh, and the dashboard starts listing matching posts on the first poll cycle.
Is this a one-time purchase or a subscription?
One-time. You pay $99 once for the software and 12 months of updates. There is no per-seat or per-lead fee from us — the only ongoing cost is your own LLM API usage, which lands at a fraction of a cent per scored post, plus any platform API keys you choose to connect.
What is your refund and update policy?
If you follow the README and the agent does not work as described, you get a full refund within 14 days of purchase — no questions asked. Updates (new platform connectors, bug fixes and prompt improvements) are included for 12 months from the purchase date.
What tech stack does it run on, and do I need a server?
It ships as a Docker image or an npm package: Node.js crawler workers, a lightweight local dashboard and SQLite storage. It runs on macOS, Windows or Linux with about 2 GB of free RAM and outbound internet access — no cloud server required, and everything stays on your machine.
Can I add platforms or tune how leads are scored, and where do I get help if I am stuck?
Yes — the prompts/ folder contains the intent and reply templates in plain text you can edit, and config.yaml exposes per-keyword and per-platform thresholds. If a community is not in the list but has a search endpoint or RSS feed, a documented generic adapter lets you add it. If you get stuck connecting a key or changing behavior, email support with your config.yaml and logs and you will get a concrete fix, not a link to the docs.
只需你的 API key
填上 config.yaml 里的 key,15 分钟跑通第一个结果
一次性买断 · 含 1 年更新 · 14 天退款保障