Anthropic recognized Databasus through their Claude for Open Source program. I want to talk about what this means, how the program works and how Databasus got here. How Databasus started In 2023 I was working on several production projects and home projects, all running PostgreSQL. Every project needed backups. And every time it was the same story: write a bash script, set up a cron job, figure out where to store the dumps, hope the notification webhook still works, repeat. Multiply that by a dozen databases across different teams and environments and it gets old fast. At some point I got tired of doing this over and over. So I built a small internal tool to handle it. Nothing fancy. Just a web UI on top of pg_dump with scheduling, a couple of storage options and email notifications when something went wrong. The idea was that anyone on the team could configure and monitor backups without touching the terminal. I called it Postgresus. It ran on our servers, backed up our production databases, and I didn't think much about it beyond that. Going open source In early 2025 I decided to open source it. The reasoning was simple: if it saves me time, maybe it saves other people time too. I pushed it to GitHub, wrote a README, set up a Docker image and shared it on Reddit and a few other places. The response was bigger than I expected. Within weeks there were hundreds of stars, then thousands. Docker pulls started climbing. People were actually using it. Not just solo developers, but DevOps teams, DBAs, startups and companies running it in production for real workloads. I started getting messages from teams who replaced their internal backup scripts with Databasus and from companies that needed something simpler than pgBackRest but still reliable enough for production. By the end of 2025, Postgresus had become the most starred PostgreSQL backup tool on GitHub, passing WAL-G, pgBackRest and Barman. These are established, enterprise-grade CLI tools that have been around for years. A web UI tool overtaking them on stars showed that there's real demand for backup solutions that teams can actually operate without deep infrastructure expertise. The rename to Databasus By late 2025 the project had outgrown its name. Three things made the rename necessary. First, "Postgres" is a trademark of PostgreSQL Inc. and you technically can't use it in project names. I'd been aware of this for a while but kept pushing it off. Second, the tool was no longer just for PostgreSQL. We had added MySQL, MariaDB and MongoDB support. Calling it "Postgresus" while it backed up MongoDB didn't make much sense. Third, the scope had changed. It started as a wrapper around pg_dump. By that point it had encryption, retention policies, team workspaces, RBAC, audit logs, health monitoring and 70+ storage destinations via Rclone. The project had evolved from a solo developer tool into something that teams and enterprises were running in production. It was a different project from what I originally built. So in the end of 2025 we renamed it to Databasus. Same tool, new name that actually fits what it does. What Databasus is today For those unfamiliar: Databasus is a free, open source, self-hosted backup management tool for databases. PostgreSQL is still the primary focus, but MySQL, MariaDB and MongoDB are fully supported. Apache 2.0 license. Runs as a single Docker container or via Helm on Kubernetes. All data stays on your infrastructure, which matters a lot for teams with strict data governance requirements. Here's what it does: Scheduled backups: hourly, daily, weekly, monthly, or custom cron expressions
70+ storage destinations: local disk, S3, Cloudflare R2, Google Drive, Dropbox, SFTP, NAS, and anything else supported by Rclone
AES-256-GCM encryption with unique keys per backup file
Retention policies: time period, count, size limit, or GFS (Grandfather-Father-Son). GFS is often a hard requirement for enterprises that need to satisfy regulatory compliance or audit obligations
Database health monitoring with configurable failure thresholds, so your team knows when something is wrong before it becomes a disaster
Team features: workspaces, role-based access control (viewer/member/admin/owner), audit logs. Built for organizations where multiple people need to manage backups across different projects
Notifications: Slack, Discord, Telegram, MS Teams, Email, webhooks
Works with cloud-managed databases: AWS RDS, Google Cloud SQL, Azure Database. This is something traditional tools like pgBackRest and Barman can't do, since they require direct filesystem access
Full backup portability: backups can be decrypted and restored using only the secret.key file and standard tools, without Databasus itself. No vendor lock-in. This matters for enterprises that need a guarantee they can recover their data even if the backup tool itself is gone Scheduled backups: hourly, daily, weekly, monthly, or custom cron expressions Scheduled backups 70+ storage destinations: local disk, S3, Cloudflare R2, Google Drive, Dropbox, SFTP, NAS, and anything else supported by Rclone 70+ storage destinations: AES-256-GCM encryption with unique keys per backup file AES-256-GCM encryption Retention policies: time period, count, size limit, or GFS (Grandfather-Father-Son). GFS is often a hard requirement for enterprises that need to satisfy regulatory compliance or audit obligations Retention policies: Database health monitoring with configurable failure thresholds, so your team knows when something is wrong before it becomes a disaster Database health monitoring Team features: workspaces, role-based access control (viewer/member/admin/owner), audit logs. Built for organizations where multiple people need to manage backups across different projects Team features: Notifications: Slack, Discord, Telegram, MS Teams, Email, webhooks Notifications: Works with cloud-managed databases: AWS RDS, Google Cloud SQL, Azure Database. This is something traditional tools like pgBackRest and Barman can't do, since they require direct filesystem access Works with cloud-managed databases: Full backup portability: backups can be decrypted and restored using only the secret.key file and standard tools, without Databasus itself. No vendor lock-in. This matters for enterprises that need a guarantee they can recover their data even if the backup tool itself is gone Full backup portability: As of early 2026, the project has around 5,800 GitHub stars, over 250,000 Docker pulls, and tens of thousands of daily active users. The user base ranges from individual developers backing up side projects to engineering teams at companies running Databasus across dozens of production databases. Anthropic's Claude for Open Source program Now to the main news. Anthropic runs a program called Claude for Open Source. It gives eligible open source maintainers 6 months of free Claude Max 20x, which is the highest-tier consumer Claude subscription with 20x the usage limits of Claude Pro. The program has two eligibility tracks. The Maintainer Track requires all of the following: you need to be a primary maintainer or core team member of a public GitHub repo, the repo needs 5,000+ GitHub stars or 1,000,000+ monthly NPM downloads, you must have been active in the repo within the last 3 months (commits, issue triage, PR reviews, releases), and you need merge/write access. There's also an Ecosystem Impact Track for projects that don't meet the star threshold but that the ecosystem meaningfully depends on. You write a short explanation of the project's significance and Anthropic evaluates it based on downstream dependents, breadth of usage and criticality. Applications are reviewed on a rolling basis. Meeting the thresholds doesn't guarantee acceptance. Anthropic approves or denies at their discretion. The total cap is 10,000 recipients, and the application deadline is June 30, 2026. It's worth noting what the program signals. The requirements are not low. 5,000 stars, active maintenance and real adoption. They're looking for projects that people actually depend on, not hobby repos with inflated metrics. Getting accepted means your project passed that filter. Databasus and the program Databasus qualified through the Maintainer Track. The star count exceeds the threshold, the project is under active development and the adoption numbers speak for themselves. For me, this was a meaningful moment. Not because of the subscription itself, but because of what it represents. A company like Anthropic looked at the project, evaluated it against their criteria and decided it's worth supporting. That's a credibility signal that matters, especially for a tool that handles something as sensitive as database backups. When teams evaluate whether to trust an open source tool with their production data, signals like this help. How AI is used in Databasus I use Claude in the development process. I want to be upfront about that because the topic of AI in open source has become contentious, and for good reason. There's a lot of low-quality AI-generated code being pushed into open source projects right now. For a tool that teams and enterprises trust with their production data, this transparency is not optional. Databasus has a dedicated AI disclaimer in the README (https://github.com/databasus/databasus?tab=readme-ov-file#ai-disclaimer) that explains exactly how AI is and isn't used. https://github.com/databasus/databasus?tab=readme-ov-file#ai-disclaimer AI is used for: AI is used for: Verification of code quality and searching for vulnerabilities
Cleaning up and improving documentation, comments and code
Assistance during development
Double-checking PRs and commits after human review Verification of code quality and searching for vulnerabilities Cleaning up and improving documentation, comments and code Assistance during development Double-checking PRs and commits after human review AI is NOT used for: AI is NOT used for: Writing entire code
"Vibe code" approach
Code without line-by-line verification by a human
Code without tests Writing entire code "Vibe code" approach Code without line-by-line verification by a human Code without tests The project has solid test coverage, both unit and integration tests. There's a CI/CD pipeline with automated tests and linting. Every PR goes through manual verification by experienced developers. Vibe coded PRs are rejected by default. And to be clear, the same applies to poor code written by humans. We don't differentiate between bad human code and AI slop. The bar is the same for everyone. There were actual cases where we denied PRs specifically because they were obviously AI-generated without proper review. You can see an example in this GitHub issue: https://github.com/databasus/databasus/issues/145 https://github.com/databasus/databasus/issues/145 My position on this is simple. AI is a tool. A good one. But you don't get to hand off responsibility for what you ship. Teams and companies trust your project with their production data. That trust has to be earned through engineering rigor, not outsourced to a language model. I think it's important for open source projects to be transparent about this. If you use AI, say how. If you don't allow vibe coding, enforce it. Especially when your users are running your code in production environments where a bad backup means real data loss. What's next Databasus keeps growing. We're working on new features, improving existing ones and the community around the project continues to expand. Incremental WAL backups with PITR are in development! Having Anthropic's support through the Claude for Open Source program is motivating and helps keep the momentum going. If you're an open source maintainer and your project meets the criteria, I'd recommend applying. The program deadline is June 30, 2026 and they review applications on a rolling basis, so earlier is better. And if your team is looking for a backup tool that doesn't require writing bash scripts or configuring pgBackRest from scratch, give Databasus a try. It takes a few minutes to set up and your team can manage everything from the web UI. GitHub: https://github.com/databasus/databasus https://github.com/databasus/databasus Website: https://databasus.com https://databasus.com Claude for Open Source program: https://claude.com/contact-sales/claude-for-oss https://claude.com/contact-sales/claude-for-oss

This story contains new, firsthand information uncovered by the writer.

This story contains AI-generated text. The author has used AI either for research, to generate outlines, or write the text itself. 

Anthropic Backs Databasus Under Claude for Open Source Program

About Author

Comments

TOPICS

THIS ARTICLE WAS FEATURED IN

Related Stories

Databasus Became the Most Popular PostgreSQL Backup Tool in 2025

Databasus Became the Most Popular PostgreSQL Backup Tool in 2025

The Noonification: Understanding How Data Warehousing on AWS Works (12/3/2023)

2019 Database Trends – SQL vs. NoSQL, Top Databases, Single vs. Multiple Database Use

47 Stories To Learn About Postgres

6 Reasons You Should Transition to PostgreSQL from Oracle

Databasus Became the Most Popular PostgreSQL Backup Tool in 2025

Databasus Became the Most Popular PostgreSQL Backup Tool in 2025

The Noonification: Understanding How Data Warehousing on AWS Works (12/3/2023)

2019 Database Trends – SQL vs. NoSQL, Top Databases, Single vs. Multiple Database Use

47 Stories To Learn About Postgres

6 Reasons You Should Transition to PostgreSQL from Oracle

Light-Mode

Classic

Newspaper

Minty

Dark-Mode

Neon Noir

Minty

HN StartUps