Skip to content

FAQ.md

Latest commit

 

History

History
94 lines (69 loc) · 4.09 KB

FAQ.md

File metadata and controls

94 lines (69 loc) · 4.09 KB
Raw

FAQ

Do I need to write SQL myself or can my AI agent do that?

While you are welcome to manually use PlyDB with your hand crafted SQL, AI agents can be very good at writing SQL too - exploring, understanding, and analysing datasets - even if your data is less organized than you'd like it to be.

Should I use MCP or CLI?

It depends. There are tradeoffs and limitations depending on your situation.

Having an AI agent use PlyDB as a CLI tool can be a more dynamic workflow. For example, AI agents can reconfigure plydb between calls — changing config files, or evolving semantic context overlays on the fly. In contrast, PlyDB's MCP server will need to be restarted to change its configuration. Usually this means restarting your agent, too. (Though we might address this in the future.)

Your choice may also be informed by what your agent's limitations are around CLI tool calling. For example, OpenClaw allows very permissive access to your system, while Claude Cowork operates in an isolated VM sandbox that strictly limits allowed tools and outbound networking, and Claude Code is somewhere in the middle.

Pro Tip: When first setting up or evolving your configuration, we recommend pairing a CLI-capable AI agent with the PlyDB CLI. This allows the agent to actively assist with data source setup and capture semantics into a context overlay - preserving its insights for future sessions. Once your configuration has stabilized, you can transition to whichever method works best for your preferred agent.

See also:

How organized should my data be?

Even if your data isn't well organized, you may be pleasantly surprised at how capable AI agents are at making sense of it. We suggest you give it a shot!

Pro tip: After a session of data analysis, ask your AI agent to distill its learnings about your data's semantics and write a semantic context overlay to record its findings for future sessions.

Can I have an AI agent write my PlyDB config file for me?

Yes! Install the PlyDB Agent Skill or point your agent at the PlyDB config schema and tell it about the data sources you want to configure.

Can I have an AI agent write my PlyDB semantic context overlays for me?

Yes! Install the PlyDB Agent Skill to teach it about PlyDB semantic context overlays.

Try asking your AI agent to write an overlay file after a data analysis session — it's a particularly good opportunity to capture learnings for future sessions.

Does the PlyDB CLI work with Claude Cowork?

Yes, but with some limitations when using PlyDB via the CLI (not MCP).

Limitations:

  • Claude Cowork's security sandboxing requires that the plydb binary be available in a directory that you've granted Claude access to. For example, you can place it in your project's workspace.
  • Claude Cowork's sandboxing also limits network access from within the sandbox. This means PlyDB is not able to connect to networked data sources or download extensions for its data connectors.

When working with local CSV files as data sources, you should not run into the networking restrictions.

However, when connecting to networked data sources such as Postgres, MySQL, S3, or Google Sheets, it is recommended that you integrate PlyDB via MCP instead of CLI when using Claude Cowork.

Does the PlyDB CLI work with Claude Code?

Yes. Claude Code does not use the same security sandboxing as Claude Cowork. Claude Code can run any tool available on your system (with your permission).