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

# Use Cases & Examples

> Real prompts and end-to-end examples for the Writesonic Agents MCP server.

Every example below is something you type to your connected assistant in plain language. The assistant picks the tools; the italic lines show the calls it makes under the hood.

## Start every session with your projects

Because almost every tool is project-scoped, the assistant resolves your project first.

> **You:** What can you do with my Writesonic Agents account?
>
> *Assistant calls `whoami` → returns your identity and projects, then works within the project you pick.*

<Tip>
  If you have several projects, name the one you mean ("in my Acme project…"). Otherwise the assistant will ask rather than guess.
</Tip>

## Triage your task list

> **You:** How many open tasks do I have, grouped by assignee?
>
> *Assistant calls `query_tasks` with `view: "counts"` and `group_by: "assignee_id"`.*

> **You:** Show me the high-priority tasks that aren't done yet.
>
> *Assistant calls `query_tasks` with `view: "list"` and a filter on priority and status.*

## Create and manage a task

> **You:** Create a task "Refresh the pricing page for AI search" and assign it to me.
>
> *Assistant calls `create_task`, then `update_task` with `action: "assign"`.*

> **You:** Move it to In Progress and add a comment with the target keywords.
>
> *Assistant calls `update_task` (`action: "transition_status"`) then `comment_on_task`.*

## Follow the work an agent did

> **You:** What did the agent do on task ABC-123? Walk me through the steps.
>
> *Assistant calls `get_agent_runs` with `task_id`, then `run_progress` for the run to show each step.*

## Inspect a playbook and its runs

> **You:** List my content playbooks and show the last few runs of the blog one.
>
> *Assistant calls `list_workflows` (`view: "list"`), then `get_workflow` (`view: "runs"`) for the matching playbook.*

## Read a grid

> **You:** What columns are in my "Q3 landing pages" grid, and show the first 20 rows.
>
> *Assistant calls `list_grids` to find it, then `get_grid_data` (`view: "columns"` then `view: "rows"`).*

## Search your knowledge and brand kit

> **You:** What does our brand voice guide say about tone?
>
> *Assistant calls `get_brand_kit` for the voice section, or `search_knowledge` across your sources.*

## Pull a content-intelligence report

> **You:** How is our /pricing page doing in AI citations this month?
>
> *Assistant calls `get_ci_dashboard` (`view: "pages"`) to resolve the page, then `get_page_report` with the AI-citations `report_type`.*

## Chain several reads in one shot

For a multi-step question, the assistant can batch the calls instead of round-tripping each one.

> **You:** Give me a status snapshot: open task counts, my three most recent playbook runs, and any failing integrations.
>
> *Assistant uses `execute_batch` to run `query_tasks`, `get_workflow` (runs), and `list_integrations` in one request.*

## Three things to take note of

* **Name your project.** Almost every tool is project-scoped, so the assistant resolves it with `whoami` first. Say which project you mean ("in my Acme project…") and it won't have to stop and ask.
* **Writes always confirm first.** Only `create_task`, `update_task`, and `comment_on_task` change anything, everything else is read-only. The assistant shows you the change and asks before writing, and each write is reversible in the product.
* **Batch multi-step questions.** For a question that needs several reads, ask for the whole thing in one message. The assistant chains the calls with `execute_batch` (or `execute_plan` when there are loops or conditionals) instead of round-tripping each one.

## Where to go next

<CardGroup cols={2}>
  <Card title="Tools Reference" icon="wrench" href="/agents-mcp/tools-reference">
    Every tool and its arguments
  </Card>

  <Card title="Agents REST API" icon="code" href="/api-reference/agents/authentication">
    Call the same tools from code
  </Card>
</CardGroup>