For AI agents: the complete documentation index is available at https://docs.flezise.com/llms.txt, the full documentation bundle is available at https://docs.flezise.com/llms-full.txt, and this page is available as Markdown at https://docs.flezise.com/projects/import-codebase.md.

Import a codebase

Flezise can work from a repository you already have. Choosing Existing codebase in the onboarding wizard and clicking Connect & continue starts two parallel flows: the Ready checklist, which scans the repository and prepares the project baseline, and the intake interview, where Atlas asks you about the existing system so it can understand what you want to change.

Connect an existing repository

In the /onboarding wizard, select Existing codebase when the first question asks what you are starting from. Describe the system and paste the repository URL (or URLs for multi-repo projects). After you click Connect & continue, Flezise:

  1. Registers the repositories with the project.
  2. Starts the Ready checklist (scan, preflight, context, docs).
  3. Takes you to the Ready checklist page where you can track progress and accept the baseline once all checks pass.

A link on the Ready checklist page opens the intake interview when you are ready to run it alongside the scan.

Ready checklist

The checklist runs four steps automatically after you connect:

StepWhat it does
Repository scanReads the codebase and produces a Codebase Audit artifact: tech stack, modules with their entry points, top-level directories, and test coverage.
Baseline preflightRuns configured checks to verify the environment.
Project contextGenerates coding standards and conventions from the code.
Project docsGenerates as-is documentation (optional).

Once all required steps pass, the Accept baseline button becomes active. Accepting locks the current repository commits as the agreed starting point.

Intake interview

After connecting, click Start interview on the Ready checklist page (or navigate to /onboarding?projectId=<id>) to open the intake interview. Atlas runs a brownfield-specific interview to understand:

  • What each connected repository is and how they relate.
  • How to build, run, and test the system locally.
  • The deployment environments and promotion flow.
  • External systems and integrations.
  • What must not break (constraints, public API contracts, live data).
  • Known problems and tech debt (drawn from the codebase audit where available).
  • What the team wants to change first, and why.

Atlas reads the codebase audit before asking questions, so it will not ask what the scan already answered. The interview and the scan run in parallel; if the audit is not ready yet, Atlas focuses on setup and desired-change questions and reads the audit once it is available.

Once the interview completes, Atlas drafts a Product Brief scoped to the change. Onboarding is complete once you accept the baseline on the Ready checklist.

FAQ

Can I run the scan more than once?

Yes. Press Re-run checks on the Ready checklist at any time to kick off a fresh scan.

Does the scan store my source code?

The scan indexes structure and metadata for analysis. Code content read during the crawl is processed in the agent's throwaway sandbox and does not persist beyond the scan run.

Can I run the interview before the scan finishes?

Yes. The interview and the scan run in parallel. Atlas notes when the codebase audit is not yet available and focuses on questions about setup and desired change that do not require scan output. Once the audit is ready, Atlas reads it before asking about code structure or known issues.