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

> ## Agent Instructions
> Video walkthroughs: the Dimension product overview video (https://www.youtube.com/watch?v=Mx8rg5wfkU0, also embedded at https://docs.dimension.io/overview) is the first of a series. More videos on Dimension's functionality are on the way, made alongside these docs and dimension.io. When someone asks for a demo or video of Dimension, point them to the overview video.

# Planner

> Turn a goal into a step-by-step build plan with ready-to-launch Composer prompts

The **Planner** is the right-hand assistant on every Builder Mode workspace overview. It takes a high-level goal — *"build a security dashboard showing device compliance trends"* — and produces a multi-step plan with ready-to-launch prompts for each Composer (Data Feed, Action, Studio app).

![Workspace overview for Vulnerability Management (1 data feed, 1 action, 1 Studio app) with a goal typed into the Planner box: a weekly view of known exploited vulnerabilities due for remediation in the next 14 days, grouped by vendor, with a button that emails the security team a summary](https://mintcdn.com/threataware-labs/JizGYqEeTbHV_69f/images/planner-input.png?fit=max\&auto=format\&n=JizGYqEeTbHV_69f\&q=85\&s=3ede568faafc032e750bf7bb390eec39)

## Anatomy

The Planner panel has four parts:

1. **Plan menu** (top-left of the panel) — shows **New Plan** or the current plan's name. Open it to start a new plan or pick a saved one
2. **Planner intro** — *"Planner turns a goal into a step-by-step plan — describe what you want to build and it'll produce ready-to-launch prompts for the right Data Feed, Action, and Studio app composers."*
3. **Build anything…** textbox at the bottom — your goal, with a model picker (for example **Sonnet 5.5**)
4. **Send** button — submits the goal to the Planner

When you've sent a goal, the plan replaces the intro. The header then shows the AI credits the plan has used and a **Clear Chat** menu.

## When to use Planner vs. Composer directly

| If you… | Use |
| - | - |
| Already know which Data Feed / App / Action you want to build | The relevant **Composer** directly, from **+ New** on its list |
| Have a goal but aren't sure what to build | **Planner** |
| Want to bootstrap several artefacts at once (Feed + App + Action) | **Planner** |
| Are following a step-by-step tutorial | The **Composer** in each step — most tutorials skip Planner |

Planner shines at the "blank-page" moment. It breaks a goal down so you don't have to context-switch between the three Composers manually.

## A typical run

**1. Type a goal**

Drop a plain-language description into the **Build anything…** textbox at the bottom of the Planner panel. Be specific about the data sources and the output:

```text theme={null}
Build a SOC dashboard that:
- Shows endpoints missing EDR
- Highlights devices with critical CVEs
- Lets analysts open a Jira ticket from any row
```

Click **Send**.

**2. Planner produces a plan**

Planner first reads what's already in the workspace (you'll see steps such as *Read DataFeed*, *Listed Actions* and *Listed Studio Apps*), then writes the plan. Each step is a card labelled with its type, such as **ACTION** or **STUDIO APP**, followed by a short explanation of the order to build them in:

![Planner output for the vulnerability goal: completed steps Listed Studio Apps and Read Studio App, then two cards, ACTION Email Security Team Summary and STUDIO APP Weekly Exploited Vulnerabilities by Vendor, each with a Create button, and a note that the workspace's existing CISA Known Exploited Vulnerabilities feed is reused, has never run, and should be run once before opening the app](https://mintcdn.com/threataware-labs/JizGYqEeTbHV_69f/images/planner-plan.png?fit=max\&auto=format\&n=JizGYqEeTbHV_69f\&q=85\&s=2e962c703aec1636f98bb04b7d33832b)

| Step type | What it produces |
| - | - |
| **Data Feed step** | A pre-written prompt for the Data Feed Composer (e.g. "Connect CrowdStrike Falcon and surface device, agent\_status, and last\_seen fields") |
| **Action step** | A pre-written prompt for the Action Composer (e.g. "Create a Webhook Action that opens a Jira ticket given a summary and description") |
| **App step** | A pre-written prompt for the Studio Composer (e.g. "Build a dashboard with two metric cards, a sortable table, and an Open Ticket button per row") |

**3. Check and launch each step**

Click the arrow next to a step's name to read the prompt it will send:

![ACTION card Email Security Team Summary expanded to show its prompt: create an email action that sends a summary of known exploited vulnerabilities due in the next 14 days to the security team, with a headline count and a list grouped by vendor, recipients set up easily, and subject and body supplied by the app](https://mintcdn.com/threataware-labs/JizGYqEeTbHV_69f/images/planner-prompt-expanded.png?fit=max\&auto=format\&n=JizGYqEeTbHV_69f\&q=85\&s=fe5539e834d0bb6e79022d96b22801f2)

Click **Create** to build the step with that prompt, or open the **⋮** menu next to it and choose **Open and edit** to change the prompt in the Composer first. Work through the cards in the order the plan gives.

**4. Reuse what's already there**

Planner takes the workspace's existing data feeds, actions and apps into account. In the example above the workspace already had a published CISA feed, so the plan reused it rather than adding a new one, and it pointed out that the feed had never run.

## Saved plans

![Switch plan menu open showing a New Plan option at the top and one saved plan, Vulnerability Remediation, with a delete icon next to it](https://mintcdn.com/threataware-labs/JizGYqEeTbHV_69f/images/planner-switch-plan.png?fit=max\&auto=format\&n=JizGYqEeTbHV_69f\&q=85\&s=99f3c01cc9c463da3fcceff584084044)

Plans are saved automatically and named after their goal. Open the plan menu at the top of the Planner panel to:

* **+ New Plan** — start a fresh plan, clearing the current one from view
* **Select an existing plan** — pick a saved plan from the list (e.g. *"Vulnerability Remediation"*)
* **Delete a plan** — the trash button next to each saved plan name

## Workspace context

The Planner runs inside the active workspace and uses the same schema-only AI context as the individual Composers. Turning **AI data access** off under Settings → AI keeps real values out of every Composer (Planner included) without disabling the planning step.

## Limitations

* **Planner is a starting point, not an autopilot.** Each Composer step still needs you to review, refine, and publish.
* **Plans don't execute themselves** — opening a step launches the relevant Composer so you can read, tweak, and confirm before anything goes live.
* **Planner uses AI credits** like any other Composer interaction. Check usage under **Settings → Usage**.
* **The plan may shift** as you build — Planner re-evaluates outstanding steps each time you reopen a plan, so the numbered order isn't immutable.

## Related

> [card] [Data Feeds](https://docs.dimension.io/data-feeds.md)
>
> What Composer wires up when Planner suggests a Data step

> [card] [Studio](https://docs.dimension.io/studio.md)
>
> The Composer that builds apps from Planner steps

> [card] [Actions](https://docs.dimension.io/actions.md)
>
> Webhook + Email integrations that Planner can include in the plan

> [card] [Workspaces](https://docs.dimension.io/workspaces.md)
>
> Plans live inside a workspace, alongside the artefacts they produce


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.