# database-skills: PlanetScale's AI Skills for Database Work in Coding Agents

> database-skills is a PlanetScale repository that provides installable skills for AI coding assistants such as Claude Code and Cursor, covering MySQL, PostgreSQL, Vitess, and Neki. Each skill is a directory with a SKILL.md file that loads database-specific guidance into the agent's context when it encounters relevant tasks.

**planetscale/database-skills** — Skills for AI agents working with databases

- Repository: https://github.com/planetscale/database-skills
- Website: https://db-skills.com
- Stars: 693 · Forks: 61
- Language: HTML
- License: MIT
- Published: 2026-09-10 · Updated: 2026-09-10 · Language: en
- Canonical page: https://hysenlabs.com/projects/planetscale-database-skills

## What database-skills Is and Who It Is For

database-skills is a repository maintained by PlanetScale that defines installable skills for AI coding assistants. A skill, in the context of the skills.sh ecosystem, is a directory containing a SKILL.md file with instructions and triggers, plus optional reference documents. When an AI coding agent receives a relevant prompt, it loads the skill's content into its context and applies the documented guidance.

The repository targets developers working with PlanetScale's database products who want their AI coding assistant to handle database-specific tasks correctly. The four skills cover MySQL/InnoDB, PostgreSQL as hosted by PlanetScale, Vitess (PlanetScale's sharded MySQL product), and Neki (PlanetScale's sharded Postgres product). The README describes each skill's trigger condition: for example, the mysql skill activates when creating or modifying MySQL tables, diagnosing slow queries, or planning migrations.

The primary language listed is HTML, which refers to the website/ directory. The skill logic itself is in Markdown files. The last push was on September 11, 2026, and the repository has no GitHub releases.

## The Four Skills and Their Scope

The mysql skill covers MySQL/InnoDB schema design, indexing, query tuning, transactions, and operations. The README lists its trigger conditions as creating or modifying MySQL tables and indexes, diagnosing slow or locking behavior, planning migrations, and troubleshooting replication and connection issues.

The postgres skill covers PostgreSQL best practices, query optimization, connection troubleshooting, and performance insights specific to PlanetScale Postgres. Its scope includes schema design, indexing, MVCC and VACUUM behavior, WAL tuning, replication, monitoring, and PlanetScale-specific features like connection pooling and CLI insights.

The vitess skill addresses Vitess's sharding layer: VSchema configuration, keyspace management, online DDL, VReplication, and MySQL-compatible scaling issues. Vitess adds concepts (VSchema, VTgate, keyspace) that have no equivalent in plain MySQL; the skill provides guidance specific to those abstractions.

The neki skill provides an overview of Neki, PlanetScale's sharded Postgres product. The README describes it as guidance for Neki-related tasks and for evaluating how to scale and shard Postgres. Neki is distinct from PlanetScale's standard Postgres offering.

## Installing database-skills for Claude Code and Cursor

The recommended installation path uses the skills.sh CLI:

```bash
npx skills add planetscale/database-skills
```

This command installs all four skills. The README links to skills.sh for documentation on the skills.sh ecosystem and how skills are loaded at runtime.

For Cursor, the README provides an alternative path: add the repository as a Cursor plugin by typing the following in Cursor chat:

```bash
/add-plugin database-skills
```

The Cursor plugin configuration is in the .cursor-plugin/plugin.json file at the repository root. Individual skills do not have separate install commands in the README; the npx skills add command installs the whole repository at once.

Contributing a new skill requires forking the repository, creating a new directory under skills/, adding a SKILL.md and optional references/ directory, then opening a pull request. The package.json lists wrangler as the only development dependency, which is used for the website/ portion of the repository hosted at db-skills.com.

## The Skill File Format: SKILL.md and References

Each skill directory under skills/ contains two things. The required file is SKILL.md, which holds the instructions, triggers, and workflow for the skill. The optional references/ directory holds supporting documents that the skill loads into context as needed.

The README's repository structure shows this layout:

```
database-skills/
└── skills/
    ├── mysql/
    │   ├── SKILL.md
    │   └── references/
    ├── postgres/
    │   ├── SKILL.md
    │   └── references/
    └── vitess/
        ├── SKILL.md
        └── references/
```

The repository does not document what the SKILL.md files contain beyond the high-level descriptions in the README. The trigger conditions listed in the README ("use when creating or modifying MySQL tables") come from those files, but their full contents are not shown in the README. The reference documents in each skill's references/ directory presumably contain more detailed database-specific guidance, configuration examples, or troubleshooting runbooks.

## What database-skills Does Not Do

The skills are purely instructional. They contain no executable code, no database connection logic, and no tooling. They do not query your database, run EXPLAIN on a slow query, or read your schema. Their value is in shaping how the AI assistant reasons and writes SQL or configuration when those tasks appear in a conversation.

This means the quality of the output depends on how well the SKILL.md files capture the relevant guidance and how accurately the agent applies it. The repository's README does not document the content of the individual SKILL.md files, so it is not possible to verify from the README alone what guidance each skill provides.

The skills are specific to PlanetScale's products and deployment model. A team using self-hosted MySQL without any PlanetScale infrastructure will get less value from the postgres and vitess skills, which target PlanetScale-specific behaviors like their connection pooling layer and PlanetScale CLI insights.

The repository has no GitHub releases and no versioning scheme visible in the README. Updates to the SKILL.md files happen as commits to the main branch. Any cached copy of the skills in a local skills.sh installation may fall behind the current version if it is not refreshed. The README does not document a refresh or update command.

## database-skills vs. Writing a Custom System Prompt

The simplest alternative to database-skills is writing a custom system prompt or CLAUDE.md file that instructs the AI assistant about your database environment. This approach requires no installation and no external dependency; the guidance lives in your project's own files.

database-skills provides a pre-written, maintained version of that guidance for the specific case of PlanetScale's MySQL, Postgres, Vitess, and Neki products. PlanetScale maintains the SKILL.md files and updates them as their products evolve. The trade-off is that the repository scope is fixed to those four technologies; a team working with CockroachDB, SQLite, or Cassandra would need to write their own skills or system prompts.

The skills.sh ecosystem is also newer and less widely documented than writing a CLAUDE.md file or a custom system prompt directly. Relying on npx skills add adds a runtime dependency on the skills.sh CLI that must be present in each developer's environment. An engineer who already has a well-tuned CLAUDE.md for their database environment may see little additional gain from installing a separate skills package, but teams starting from scratch will benefit from the curated guidance without having to author it themselves.

## Conclusion

database-skills is worth adding to Claude Code or Cursor workflows that involve PlanetScale's MySQL, Postgres, or Vitess products. It provides guidance that is more specific than a generic SQL prompt, particularly for Vitess VSchema configuration and PlanetScale-specific Postgres features like connection pooling. It is not a substitute for reading the full PlanetScale documentation on complex migrations or VReplication setups. Before installing, verify that the skills.sh CLI is available in your environment and check whether the Cursor plugin path matches your Cursor version.

## FAQ

### How do I install database-skills for Claude Code?

Run npx skills add planetscale/database-skills in your terminal. This installs all four skills (mysql, postgres, vitess, neki) using the skills.sh CLI. The README links to skills.sh for documentation on how skills are loaded during a Claude Code session.

### Does database-skills work with Cursor?

Yes. The README describes a Cursor plugin installation path: type /add-plugin database-skills in Cursor chat. The plugin configuration is in the .cursor-plugin/plugin.json file in the repository.

### What databases does database-skills cover?

The repository includes four skills: mysql (MySQL and InnoDB), postgres (PostgreSQL as hosted by PlanetScale), vitess (PlanetScale's sharded MySQL product), and neki (PlanetScale's sharded Postgres product). The skills are tailored to PlanetScale's hosting environment, not general-purpose database deployments.

## Sources

- [Issues](https://github.com/planetscale/database-skills/issues)
- [License: MIT](https://github.com/planetscale/database-skills/blob/main/LICENSE)
- [planetscale/database-skills on GitHub](https://github.com/planetscale/database-skills)
- [Project website](https://db-skills.com)
- [README](https://github.com/planetscale/database-skills/blob/main/README.md)

---

Hysen Labs editorial analysis, written from the project's own repository and release notes. Cite the canonical page: https://hysenlabs.com/projects/planetscale-database-skills
