Bill Nothing by Accident With a Grok Bot Template for Lead Finder Runs

The scary part of pointing an agent at a Grok bot template for lead finder work is not bad keywords, it is billing. An agent that can add tracked keywords can also add cost. RedReplier's API is built so it cannot do that in one step, and understanding that two-step is the difference between a useful agent and one you have to supervise constantly.

The short version: keywords added through the API land as PENDING and monitor nothing. GET /keywords/billing-preview prices them, and only POST /keywords/activate-pending turns them on after you say yes.

What is a Grok bot template for lead finder work?

A Grok bot template for lead finder work is a share link that copies a configured Grok Bot, plus a skill documenting the RedReplier REST API, into your own Grok Bot app. SpaceXAI shipped template sharing on 28 August 2026. Your token and your tracked keywords stay yours; only the instructions travel.

It is the lead-generation companion to the Grok bot template for social listening guide.

Why do keywords start as PENDING?

RedReplier's KeywordStatus enum is PENDING, ACTIVE, DISABLED and SUSPENDED, and anything added through the API begins at PENDING. A pending keyword is recorded but monitors nothing and bills nothing. That gap is deliberate, and it is what makes an agent safe to hand keyword suggestions to.

So a session looks like this. The bot proposes twenty keywords, prices them, and stops:

You:  Suggest keywords for a project management tool aimed at agencies,
      then price them.
 
Bot:  POST /websites/analyze-description   -> reads the site, drafts a profile
      POST /websites/{id}/keywords         -> 20 keywords added as PENDING
      GET  /keywords/billing-preview       -> "20 pending, $X at next renewal"
      "Twenty keywords are pending and cost nothing yet. Activate them?"
 
You:  Activate the twelve about agency workflows only.
 
Bot:  POST /keywords/activate-pending      -> only those twelve go ACTIVE

Nothing bills until the last line, and the last line only runs because you said so.

How does the bot find actual buying signals?

The bot filters mentions to the sources and relevance buckets where people ask for recommendations, then explains why each one scored. Reddit comments carry most of the recommendation traffic, so a lead run usually starts there and widens.

You:  Today's lead candidates: Reddit comments and Hacker News, HIGH and above,
      sorted by relevance. Explain the top five.
 
Bot:  GET  /mentions?sources=REDDIT_COMMENT,HACKERNEWS
           &scoreBuckets=HIGH,VERY_HIGH&statuses=NEW&sort=RELEVANCE
      POST /mentions/{id}/explain    (x5)

POST /mentions/{id}/explain is what turns a score into something you can act on. It returns the reasoning behind the number, so a borderline thread gets judged on its content rather than on a bucket label.

How do I install and test it?

Installing the RedReplier Grok bot template takes four steps: open the share link in Grok Bot, create a redreplier_ API token, paste it into the credential field, and ask for a keyword billing preview as a read-only test.

Can the bot manage keywords over time?

Yes, and the read-only half is where it earns its keep. GET /keywords/change-usage shows how keyword usage has moved, so the bot can flag keywords that stopped producing mentions and propose disabling them.

Turning one off is POST /keywords/{id}/disable, and re-enabling is POST /keywords/{id}/enable. Both are reversible, unlike deletion, which is why the skill prefers them.

Should the bot run lead finding on a schedule?

Run it on a schedule for reading, never for activating. A daily lead digest is useful; a routine that adds and activates keywords on its own is a billing surprise waiting to happen.

Name:      Daily lead candidates
Schedule:  every weekday 08:30
What:      GET /mentions for new REDDIT_COMMENT and HACKERNEWS mentions since
           the last run, HIGH and VERY_HIGH only, sorted by relevance.
           Summarise the top five with a one-line reason and a direct link.
Approval:  never activate keywords and never change a mention status.
If empty:  do nothing, send no notification.

What breaks in the first week?

The most common one looks like a broken integration and is not: a keyword was added and never activated, so it monitors nothing. Ask the bot for the billing preview and you will see it sitting in pending.

After that, a 401 on the first call is usually the plan rather than the token, and a lead run filtered to REDDIT_POST alone will miss the comment threads where recommendations actually happen. The Reddit monitoring guide covers that split in detail.

Frequently asked questions

Can the Grok bot spend money without asking me?

No. Adding keywords costs nothing because they land as PENDING. Only POST /keywords/activate-pending starts billing, and the skill instructs the bot to show the billing preview and wait for an explicit yes before calling it.

What does the bot use to suggest keywords in the first place?

POST /websites/analyze-description reads a URL and drafts a description of what the site sells, which the bot then turns into keyword candidates. You can also just tell it what you sell and skip that call.

Can it find leads on Hacker News as well as Reddit?

Yes. HACKERNEWS is one of the six sources, alongside REDDIT_POST, REDDIT_COMMENT, TWITTER, BLUESKY and FACEBOOK. For B2B and developer tools, Hacker News plus Reddit comments is usually the highest-signal pair.

Does disabling a keyword delete its history?

No. POST /keywords/{id}/disable stops future monitoring and leaves existing mentions in place, which is why it is the reversible option the template prefers over deletion.

Where do I find other Grok bot templates?

groktemplates.dev indexes Grok bot templates by use case. On this blog, the Reddit monitoring and X social listening guides cover the other two jobs this template does.

Install the RedReplier Grok bot template and run a billing preview before anything goes live, or read the API docs first.

Sources: the RedReplier OpenAPI spec, x.ai/bot, and @bot on X (28 August 2026). Checked 30 August 2026.