From the first stable release to GitHub Packages, optional evidence-grounded AI, a web container image and a better mobile experience — this is the RepoDNA journey from v1.0.0 to v1.2.0.
Cover image
docs/media/promo/launch-16x9.png
Tags
opensource rust devtools softwareengineering
Article
On September 27, 2026, RepoDNA reached its first stable release.
On October 5, 2026, it reached v1.2.0.
In between, the project went through four releases that each added another layer to the same idea:
Understand your codebase. See its DNA.
This is the story of that progression.
v1.0.0 — The foundation
RepoDNA v1.0.0 was the first stable release of the project.
The goal was to build an evidence-first repository intelligence and code archaeology platform that could answer questions developers routinely face when entering or maintaining a codebase.
How is this organized?
What depends on what?
Where is the architecture?
What changed over time?
Where are the hotspots?
What deserves a closer look?
The first stable release brought together analysis of structure, languages, architecture, dependencies, Git history, quality, security, tests, builds and documentation.
It also introduced the Codebase Time Machine, Project DNA cards, reports, comparisons, onboarding guides and portable .repodna artifacts.
The same release shipped the CLI, local web interface, desktop app and container image.
Most importantly, RepoDNA established its design principles:
Evidence first.
Local-first.
No telemetry.
Descriptive rather than judgmental.
Honest about limitations.
Release notes:
https://github.com/sanskarIN/RepoDNA/blob/main/docs/releases/v1.0.0/README.md
v1.1.0 — Adding AI without replacing the analysis
The next major step was v1.1.0.
This release added optional AI explanations.
But rather than handing an entire repository to a model and asking for a summary, RepoDNA takes a different approach.
The model receives a numbered selection of evidence from the analysis.
That evidence can cover architecture, history, dependencies, onboarding, modules, hotspots or a question you ask.
The resulting answer contains provenance information such as the provider, model, revision, cited evidence and token usage.
RepoDNA then checks the response.
Unsupported citations can be removed.
Unsupported statements are identified.
Inferences are labeled as inferences.
AI also remains off until the user configures a provider.
Remote AI requires explicit permission, API keys are kept outside the configuration, and source excerpts are not sent unless the user explicitly enables them.
There is even a dry-run mode:
repodna explain --dry-run
This lets you inspect what would be sent without actually contacting a provider.
So v1.1.0 did not change the fundamental philosophy of RepoDNA.
It added an optional explanation layer on top of the evidence-first analysis.
Release notes:
https://github.com/sanskarIN/RepoDNA/blob/main/docs/releases/v1.1.0/README.md
v1.1.1 — Reliability and polish
v1.1.1 was a maintenance release.
The headline features were smaller, but they addressed many details that matter when software is used regularly.
CLI output became safer for shell scripting.
Local AI behavior improved for processes that keep output streams open.
Repositories without inferred architectures received clearer output.
Singular counts were corrected.
Onboarding and first-look answers were refined.
The Architecture view became clearer.
The web interface and reports received additional fixes.
The documentation was also updated to explain network behavior, GitHub Pages behavior and security naming more accurately.
And the project began making the creator's other open-source work easier to discover directly from RepoDNA.
Release notes:
https://github.com/sanskarIN/RepoDNA/blob/main/docs/releases/v1.1.1/README.md
v1.2.0 — Easier distribution and a better web experience
Then came v1.2.0.
This release expands how RepoDNA can be distributed and hosted.
A new web-interface container image is available:
ghcr.io/sanskarin/repodna-web:1.2.0
The project can now also publish packages through GitHub Packages:
@sanskarin/repodna
@sanskarin/repodna-schema
@sanskarin/repodna-visualization
That makes RepoDNA easier to integrate into JavaScript and TypeScript workflows.
The web interface also received a substantial responsive-design pass.
Phone and narrow-window behavior now handles:
- navigation
- long commands
- long URLs
- reports
- tables
- tooltips
- search
- chart controls
The release also introduces copy buttons for commands throughout the interface and onboarding experience.
Table sorting was improved so numeric filename ordering behaves naturally.
Search now communicates its result limits more clearly.
Reports wrap long paths rather than forcing a phone-sized screen to become effectively desktop-width.
This is a different kind of milestone from v1.0.0.
The first stable release established what RepoDNA was.
v1.2.0 focuses on making it easier to obtain, run and use.
Release notes:
https://github.com/sanskarIN/RepoDNA/blob/main/docs/releases/v1.2.0/README.md
Four releases, one direction
Looking at the releases together makes the evolution easier to see.
v1.0.0 established the core repository intelligence platform.
v1.1.0 added evidence-grounded optional AI explanations.
v1.1.1 improved reliability, clarity and polish.
v1.2.0 expanded distribution and improved the experience across modern screen sizes.
The project changed substantially, but its central idea did not.
RepoDNA is still about understanding software through the evidence already present in the repository.
Not replacing developers.
Not turning a codebase into an unexplained score.
Not requiring AI.
Not requiring a cloud account.
Instead, it tries to make the repository itself easier to understand.
Try RepoDNA
Repository:
https://github.com/sanskarIN/RepoDNA
Web version:
https://sanskarin.github.io/RepoDNA/
All releases:
https://github.com/sanskarIN/RepoDNA/blob/main/docs/releases/README.md
The latest release:
https://github.com/sanskarIN/RepoDNA/releases/tag/v1.2.0
What's next?
v1.2.0 is not the end of the project.
RepoDNA's roadmap already points toward deeper language analysis, stronger import resolution, pull-request analysis in CI, an official GitHub Action and more installation options.
The foundation is now in place.
The next challenge is making repository understanding even more useful without sacrificing the principles that made the project worth building in the first place.
Understand your codebase. See its DNA.
Top comments (0)