Your first analysis

Pick a branch, documents and language, then follow the run from queue to generated docs.

Before you start

You need a repository connected through the GitHub App, and the branch you want to analyze must exist on the remote. Analyses are metered by plan: Hobby includes 20 analyses per month, Pro includes 500.

Choose documents

Open the repository and press Analyze, then pick what to generate:

Type	What it produces
`README`	Overview, setup and an architecture summary
`ARCHITECTURE`	Modules, data flow, dependency map and risks
`API`	Endpoint reference built from detected routes
`CONTRIBUTING`	Workflow, conventions and review rules
`CHANGELOG`	Diff-driven release history
`CODE_DOC`	Per-file docs for the files you select

dxnx analyze start requests README, ARCHITECTURE and CODE_DOC by default. In the dashboard, README is preselected and you choose the other repository-level types; CODE_DOC covers the files selected for the run.

Choose a language

Documentation is generated in English or Russian. The choice applies to the whole run.

Optional scope

You can narrow a run to a list of files and add free-form instructions describing what the documentation should emphasize. Both are optional; leaving them empty analyzes the repository as a whole.

Follow the run

The analysis is queued in Trigger.dev. The dashboard shows the current status and progress for the repository, and the CLI reports the same run:

dxnx analyze status owner/repo

Behind the scenes a temporary clone is created, parsed with Tree-sitter, passed through Sentinel → Mapper → Architect → Writer, and deleted when the run finishes.

Where the results land

Generated files are stored in the workspace and shown in the Docs tab, separated by type. Their materialized paths are:

Type	Path
`README`	`README.md`
`ARCHITECTURE`	`docs/ARCHITECTURE.md`
`API`	`docs/API.md`
`CONTRIBUTING`	`CONTRIBUTING.md`
`CHANGELOG`	`CHANGELOG.md`
`CODE_DOC`	`docs/code/<source-path>.md`

Per-file docs

CODE_DOC can also be produced outside a full analysis. Run the sequence in this order:

dxnx analyze audit src/foo.ts
dxnx docs generate src/foo.ts
dxnx docs pin src/foo.ts

dxnx analyze audit creates the audit cache for the file, dxnx docs generate prints the generated preview, and dxnx docs pin stores it as permanent documentation.

Re-run and history

Pressing Analyze again starts a new run against the current branch. Past runs are listed in the dashboard and with:

dxnx analyze history

History is kept for 7 days on Hobby, 90 days on Pro and unlimited on Team.

If it fails

The branch is not found — check the branch name and that it exists on the remote.
The repository cannot be read — confirm the GitHub App installation still has access to it.
The plan quota is exhausted — check the analyses used this month on your plan.

For machine-readable details, add --json:

dxnx analyze status --json

Your first analysis

On this page