FAQ // Early Access

FAQ

What kinds of duplicate code does Dedup catch?

Dedup looks for copied code, renamed copies, lightly edited near misses, and semantic duplicates where new code performs the same job as code that already exists.

Who is Dedup for?

Dedup is for engineering teams that care about DRY code, especially teams using AI coding agents. Agents can move quickly, but they do not always know that a helper or pattern already exists elsewhere in a large codebase.

What languages are supported?

Tier 1 language support:

• TypeScript
• Rust
• Go
• Python

Tier 2 support (broader language coverage with extra indexing support):

• JavaScript
• Java
• Ruby

What does the free tier include?

Free includes 5 pull request analyses per month per repository. Pro is coming soon at $10 per month with 1000 pull request analyses per month per repository.

What data does Dedup read?

Dedup reads the repository and pull request context needed to compare new code with existing code. That includes code snippets, diffs, file paths, line ranges, and commit metadata used during review. See the Privacy Policy for storage and retention details.

What infrastructure powers inference?

Our primary infrastructure is Scaleway (France), where production API, worker, database, and storage systems run. We may use OpenRouter as external fallback infrastructure for availability. When OpenRouter is used, ZDR is always enabled for those requests. Learn more about OpenRouter ZDR.

Will Dedup spam my pull requests?

No. Dedup publishes a GitHub check run for visibility and only leaves inline comments when a finding is useful for a reviewer to act on.

Can my repository data be removed?

Yes. Email hello@dedup.bot from an address tied to the relevant GitHub organization or repository. Include the repository name and, if available, the GitHub App installation details so we can find and remove the right Dedup records.