# Getting Started

The fastest way to understand v0.6 Experience Preview is to create a real local project, land in Cockpit, inspect setup progress, and run one repair or export loop. The demo remains available as a secondary path.

## 1. Open the app

Launch Codex Research Stack. If no recent project is available, the app opens **Projects Launchpad** instead of auto-opening a demo. Choose **New project** for the primary flow, **Open existing project** for a folder that already contains `.research-console/`, or **Open demo project** only when you want the guided sample.

## 2. Read Cockpit first

After a project opens, start in **Cockpit**. Check:

- Start Here / Setup Progress;
- Research Now;
- Gate Board;
- Memory Ledger;
- Recovery Queue;
- Run Timeline;
- AI Suggestion Source;
- Export Snapshot.

This page shows the current research state and the next safe repair action.

## 3. Create a real project

Choose **New project**, then complete the four launch steps:

- project folder, project name, and one research question;
- research sources and method direction;
- optional first material draft;
- optional first candidate claim.

The app creates `.research-console/` as the source of truth for project state. Initial materials and claims remain draft or unsupported objects. They do not become usable evidence and do not pass gates automatically.

## 4. Plan an Autopilot route

Open **Research Autopilot**, enter a research goal, answer the method wizard, confirm a primary route and method modules, then generate the research route. The app creates stages, steps, gates, method requirements, and an audit log. AI suggestions remain reviewable cards until you accept or reject them.

## 5. Add materials

Open **Materials** and save any source clue:

- DOI;
- URL;
- file path;
- dataset;
- platform or social-media material;
- policy report;
- fieldnote;
- manual note.

Materials are intentionally easy to save. A material is not formal evidence yet.

## 6. Create candidate claims

Open **Claims** and write one statement you may later use in a paper or report. Unsupported claims are allowed at this stage, but they are visibly marked as unsupported.

## 7. Upgrade evidence

Open **Evidence**, select a material, and add:

- access time;
- verification status;
- rights or ethics note;
- optional DOI.

Evidence with missing fields can be saved as a draft, but it cannot support formal writing.

## 8. Review method artifacts

Open **Research Autopilot** and use the method artifact panel. Missing artifacts block the method gate. Draft or needs-revision artifacts keep the gate in revise state. Only user-confirmed accepted artifacts satisfy that method requirement.

## 9. Bind evidence to claims

In **Evidence**, bind evidence to a claim and explain how the evidence supports the exact statement.

## 10. Run checks and recovery

Open **Quality** or use Cockpit's **Cross-stage review**. The app returns repair actions instead of internal error messages. Recovery actions can be accepted, marked done, dismissed, or exported as local task cards.

## 11. Run diagnostics and export status

Open **Settings** to run a project health check or export a redacted diagnostic snapshot. Open **Export** and generate a project status file, or use **Research Autopilot** to export an Autopilot run report. A project with blockers can still export a truthful status file. Privacy or truthfulness risk blocks export.

## Optional Tools

Python, Codex, Zotero, Obsidian, internet access, and API providers are optional enhancements. Missing optional tools should not block the local project workflow.