> For the complete documentation index, see [llms.txt](https://planetz.gitbook.io/planetz/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://planetz.gitbook.io/planetz/core-ideas/overview.md).

# Overview & mental model

> Status: Core ideas · Target version 0.1.x

## What you'll learn

The handful of pieces that make up Planetz, and — more importantly — how they fit into a single loop. Once you have this mental model, every screen in the product is just a window onto one of these parts.

***

## One loop, five parts

Planetz is not a chat box. It's a loop you run a *fleet* of agents through, with your intent held safely outside the blast radius.

```mermaid
flowchart TD
    A["You — state intent<br/>(what you want / what's out of scope)"] --> B["Harness (orbit)<br/>picks the right workflow automatically"]
    B --> C["Fleet of agents<br/>run in parallel lanes, isolated in Git"]
    C --> D["Intent Ledger<br/>records every decision + its provenance,<br/>traced against your intent"]
    D --> E{"You react to outcomes<br/>(human terms: wrong / undo / good)"}
    E -->|ratify| F["Decision locked — agents can't silently rewrite it"]
    E -->|reverse / undo| G["Workspace rebuilt from intent"]
    F --> A
    G --> A
```

The five parts:

1. **Tasks & the fleet.** You queue real work as tasks and run a squad of agents in parallel lanes — directing, not typing every line. → [Multi-agent fleet](/planetz/core-ideas/multi-agent.md)
2. **The harness (orbit).** Each task runs on a governed workflow, and Planetz routes it to the right one automatically. → [The harness](/planetz/core-ideas/harness-governance.md)
3. **The Intent Ledger.** Every decision a run makes is recorded with its *provenance* and traced against what you decided. Drift becomes visible. → [The Intent Ledger](/planetz/core-ideas/intent-ledger.md)
4. **Disposable Git workspaces.** Runs are isolated, diffable, and recoverable; code is treated as a regenerable derivative of intent. → [Git integration](/planetz/core-ideas/git-integration.md)
5. **Where it runs.** Inference can stay on your machine via edge models, and scale to Cloud when you need it. → [Edge AI & data sovereignty](/planetz/core-ideas/edge-ai.md)

## The one idea behind all of it

Most tools protect the *running code* and ask you to approve every step. Planetz flips both:

> **Protect the&#x20;*****intent*****, not the code.** Keep your intent as the durable, protected asset — outside the workspace — and let the code be cheap to destroy and rebuild. Then you don't pre-approve work; you **react to outcomes** after they happen, and lock the decisions you agree with so no future run can quietly undo them.

This is why the same object — the [Intent Ledger](/planetz/core-ideas/intent-ledger.md) — is both the thing that **kills drift** and the thing that lets **non-experts** keep an AI-built system on-spec. Everything else in Planetz exists to feed and serve that loop.

## How the screens map to the parts

| You want to…                      | Go to          | Underlying part   |
| --------------------------------- | -------------- | ----------------- |
| Queue and watch work              | Task deck      | Tasks & fleet     |
| Shape an idea before running it   | Conversation   | Tasks             |
| Author and trace intent           | Spec Studio    | Intent Ledger     |
| Review what agents decided        | Decisions      | Intent Ledger     |
| See/define the process            | Workflows      | Harness           |
| Follow a run                      | Logs & summary | Harness + fleet   |
| Choose models, edge, integrations | Settings       | Edge AI + harness |

## Next

* [The Intent Ledger](/planetz/core-ideas/intent-ledger.md) — start here; it's the core innovation.
* [The harness: govern, don't pre-approve](/planetz/core-ideas/harness-governance.md) — how runs are kept safe without rubber-stamping.


---

# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.

## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter:

```
GET https://planetz.gitbook.io/planetz/core-ideas/overview.md?ask=<question>
```

The question should be specific, self-contained, and written in natural language.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.