> ## Documentation Index
> Fetch the complete documentation index at: https://docs.readdy.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Agents

> Use Agents to run SEO checks, diagnose rankings, plan content, monitor sites, and automate recurring work

Agents helps you run an agency with an AI teammate that checks SEO, diagnoses ranking problems, plans content, watches every published site's numbers, and runs recurring work on a schedule. This guide walks through each page in the Agents sidebar.

<CardGroup cols={2}>
  <Card title="Tasks & chat" icon="message" href="#tasks--chat">
    Describe what you need in plain language — each request becomes a Task with a live, streaming report.
  </Card>

  <Card title="Dashboard" icon="gauge" href="#dashboard">
    Every published site at a glance: yesterday's visitors, form submissions, leads, and appointments.
  </Card>

  <Card title="SEO toolkit" icon="magnifying-glass-chart" href="#seo-check">
    Four one-click checks: full SEO audit, local SEO, ranking diagnosis, and content planning.
  </Card>

  <Card title="Scheduled Tasks" icon="clock" href="#scheduled-tasks">
    Run tasks daily, weekly, or monthly — or chain them so one task follows another automatically.
  </Card>
</CardGroup>

## Tasks & chat

Click **New Task** (or the **+** button) and tell the agent what you need — the composer asks *"How can I help your agency today?"*. Starter chips give you one-tap prompts like **Check my SEO**, **Why am I not ranking?**, and **Plan my content**.

Each request opens a **Task**: a chat session where the agent works and streams its findings back to you. Your tasks are listed in the sidebar so you can return to any of them.

<Tip>
  Tasks keep running in the background. A pulsing dot next to a task in the sidebar means the agent is still working on it — you can browse other pages and come back without interrupting anything.
</Tip>

A few things you may see inside a task:

* **Follow-up questions** — the agent may ask you to choose between options before continuing.
* **Connect Google Search Console** — some checks read your Google Search Console data (rankings, clicks, impressions). If your site isn't connected yet, a **Connect** banner appears right in the chat; click it, authorize with Google, and the task picks up from there.
* **Retry** — if the connection drops mid-task, a Retry button reruns the last step without retyping anything.

To delete a task, hover over it in the sidebar and use the **⋯** menu.

## Dashboard

The **Dashboard** shows your whole portfolio in one screen.

### Site cards

Every published site gets a card with **yesterday's numbers**:

| Metric           | Shown                             |
| ---------------- | --------------------------------- |
| Visitors         | Always, with a day-over-day trend |
| Form Submissions | Always                            |
| Chatbot Leads    | When there are any                |
| Appointments     | When there are any                |

Sites with an SEO health score show it as a badge on the card.

### Per-site view

Click any site card to open its mini dashboard — the same four metrics plus the site's recent leads. Each stat deep-links into the matching Readdy page (Analytics, Forms, Leads, or Appointments), so drilling down is one click.

### To review

The **To review** chip in the top-right corner collects anything the agent has prepared that needs your sign-off before it takes effect. When the counter lights up, open the panel, look over each card, and **Approve** or **Reject** it. Nothing goes live on your sites without your approval.

The Dashboard also has the New Task composer at the top, so you can start a task without leaving the page.

## SEO Check

**SEO Check** runs a full SEO health audit for any of your sites:

<Steps>
  <Step title="Pick a site">
    Choose one of your sites from the dropdown (or type any domain).
  </Step>

  <Step title="Run check">
    The report streams into a new Task. The agent crawls the site, detects its business type, and runs the specialist checks.
  </Step>

  <Step title="Read the report">
    You get a **0–100 health score** with per-category scores, covering technical SEO, on-page, content quality, schema, Core Web Vitals, AI-search readiness, and images — plus a qualitative read on E-E-A-T and whether AI engines can quote your content.
  </Step>

  <Step title="Act on the plan">
    Fixes are grouped **Critical / High / Medium**, each with its business impact in plain language and a specific fix. The agent then offers to apply the fixes for you by updating the site with Readdy.
  </Step>
</Steps>

## Local Check

**Local Check** is for businesses that live or die by Google Maps and the local pack. Pick a site and run it; the agent identifies the business (name, category, service area) from the site, then assesses:

<AccordionGroup>
  <Accordion title="Google Business Profile" icon="store">
    Category fit, claimed status, star rating, and review count — it flags fewer than \~10 reviews or a rating under 4.0.
  </Accordion>

  <Accordion title="NAP consistency" icon="address-card">
    Whether the business name, address, and phone match across the site, its LocalBusiness schema, and listings.
  </Accordion>

  <Accordion title="On-page local signals" icon="location-dot">
    LocalBusiness schema completeness, opening hours, tap-to-call, and an embedded map.
  </Accordion>

  <Accordion title="Nearby competitors" icon="users">
    Who ranks in the local pack for the core service, and how the business compares on reviews.
  </Accordion>
</AccordionGroup>

Like the SEO Check, results come back as a prioritized Critical / High / Medium plan, and the agent offers to apply the on-page fixes with Readdy.

## Ranking Fix

**Ranking Fix** answers one question: *why isn't my page ranking for this keyword?*

Enter a site and a target keyword (e.g. "emergency plumber near me"). The agent reads the current Google results **backwards** — what page type, content shape, length, headings, schema, and media Google is actually rewarding for that keyword — then compares your page against the top results, detects page-type mismatches, and scores your page from the perspective of likely searchers.

The report tells you plainly why the page isn't ranking and exactly what to change, grouped Critical / High / Medium, with an offer to update the page for you.

<Tip>
  Ranking Fix works best for one page × one keyword. If you want the whole site checked, run **SEO Check** instead; if you want to know what pages to create, use **Content Plan**.
</Tip>

## Content Plan

**Content Plan** turns one seed keyword into a complete content architecture:

1. Enter a seed keyword (e.g. "driving lessons"). Optionally pick a site so internal-link suggestions use your real pages.
2. The agent expands it into related keywords and groups them by **actual Google SERP overlap** — not text similarity — into a hub-and-spoke structure.
3. You get one **pillar page** plus its **supporting pages**, each with a target keyword, search intent, and a one-line angle, plus the internal linking map showing which pages link to which.

The result is a concrete build list, and the agent offers to create the pillar and supporting pages on your site with Readdy.

## Skills

The **Skills** page shows everything your agent knows how to do, in two tabs.

### Built-in

Platform skills maintained by Readdy — for example **Site Management**, which lets the agent create, edit, preview, and publish website projects on Readdy. These are read-only and always up to date.

### My Skills

You can teach the agent your own repeatable work — a report format, a checklist, a workflow you run often:

<Steps>
  <Step title="Create a skill">
    Click **New** and fill in three fields: a name, a short description, and the instructions the agent should follow.
  </Step>

  <Step title="The agent picks it up">
    In new tasks, the agent uses your skill's **description** to decide when it applies, then follows your instructions.
  </Step>

  <Step title="Edit or delete anytime">
    Use the pencil and trash icons on each skill. Deleting asks for confirmation first.
  </Step>
</Steps>

Limits to keep in mind:

* Up to **20 custom skills** per account.
* Descriptions are capped at **200 characters**.
* Custom skills are private to your account.

<Tip>
  Write the description like a trigger condition, not a summary. "Use when I ask for a monthly traffic summary for a client site" triggers far more reliably than "traffic helper".
</Tip>

## Scheduled Tasks

The **Scheduled** page puts your agent on autopilot — every task here runs on its own, without you opening the app.

### Creating a task

Click **New Task** and fill in the form:

* **Task Name** — e.g. `daily-seo-check`.
* **Trigger** — how the task starts (see below).
* **Site** — optionally scope the task to one site so the agent doesn't have to figure it out.
* **Instructions** — what the agent should do each run.

<Tabs>
  <Tab title="On a schedule">
    Pick a frequency — **daily**, **weekdays**, **weekly** (choose one or more days), or **monthly** (choose the date) — and a time of day. Times are shown and entered in **your local timezone**.

    Power users can switch to **Advanced (cron)** and enter a raw 5-field cron expression instead — note that cron expressions are evaluated in **UTC**, with 1-minute minimum granularity.
  </Tab>

  <Tab title="After another task">
    Chain tasks: pick a predecessor from the **Run after** dropdown, and this task starts automatically when the selected task **finishes successfully**. You need at least one scheduled task before you can chain another after it.
  </Tab>
</Tabs>

<Note>
  A task's trigger can't be changed after creation. To switch a task from a schedule to a chain (or vice versa), create a new task.
</Note>

### Chains

Chained tasks show an **↳ after …** badge in the list. Chaining lets you build multi-step routines without guessing at timing:

```text theme={null}
09:00  Run SEO audit          (on a schedule)
  ↳    Publish a blog post    (after: Run SEO audit)
  ↳    Summarize the results  (after: Publish a blog post)
```

A chained task runs only when its predecessor **succeeds** — if a step fails, the chain stops there, so you never get half-finished follow-ups.

### Managing tasks

* **Enable / disable** — every task has a toggle, in the list and in the task detail. Disabled tasks stay configured but don't run (and don't trigger anything chained after them).
* **Run history** — open a task to see its past runs and results.
* **Delete** — stops the task permanently; its run history is kept.

## FAQ

<AccordionGroup>
  <Accordion title="Will the agent change my site without asking?">
    No. Anything the agent prepares that would change your sites lands in the Dashboard's **To review** queue, and only goes live after you approve it.
  </Accordion>

  <Accordion title="Why do scheduled times look different for a teammate?">
    The simple schedule picker displays times in each viewer's local timezone. The underlying schedule is the same moment for everyone. (Advanced cron expressions are always UTC.)
  </Accordion>

  <Accordion title="Why didn't my chained task run?">
    Chained tasks run only when the predecessor finishes **successfully**. Check the predecessor's last run — if it failed or is disabled, the chain stopped there.
  </Accordion>

  <Accordion title="What period do the Dashboard numbers cover?">
    Site cards show **yesterday's** visitors, form submissions, chatbot leads, and appointments, with a day-over-day trend on visitors.
  </Accordion>

  <Accordion title="A task asked me to connect Google Search Console — why?">
    Checks that read real search performance (rankings, clicks, impressions) need access to your Search Console. Click the **Connect** banner in the chat, authorize with Google, and rerun — the connection is remembered for that site.
  </Accordion>

  <Accordion title="Can I close the app while a task is running?">
    You can freely switch to other pages — running tasks keep going, marked by a pulsing dot in the sidebar. Long-running work like scheduled tasks runs entirely on our servers, whether or not the app is open.
  </Accordion>
</AccordionGroup>

## Need Help?

If this guide doesn't resolve your issue, [submit a support ticket](https://readdy.ai/feedback). We're here to help!