Who it is forA ops or research lead with a JSON dataset that should appear as a tracked database in the team workspace.
The problemAdding hundreds of records to a database one at a time through a chat is impractical.
The outcomeA database in the workspace filled with the imported rows, with views you can filter and share.
What you need first
- A Linux, macOS or Windows machine or server to run the binary, or a machine with Docker
- A web browser to open the address the installer prints
- An MCP-capable client such as Claude, ChatGPT, Codex, Cursor or Gemini CLI, if you want agents to work in the workspace
- Files to import (Markdown, CSV or a Notion export), if you are migrating content
Steps
- Connect your MCP client to the /mcp endpoint of your instance, as described in the README.
- Ask the agent to create a database with the properties your JSON needs.
- Point the bulk import at your JSON source so the rows load without the content passing through the agent.
- Open the database, add a table view with filters, and check the row count against the source file.
Prompt for your agent
Using the salt.md repo at https://github.com/saltmd/salt.md, create a database in the workspace at <instance-url> and bulk import the JSON file at <path-to-json>. Report the number of rows created and confirm the count matches the source.
Caveat. The README does not document the JSON format the bulk import expects, so inspect the import result before relying on the data.
Use this in your AI agent
Ask your agent
Using the Deduped MCP, get the setup steps for saltmd/salt.md and "Load a JSON dataset into a workspace database"
Add Deduped to Claude Code once
claude mcp add --transport http deduped https://dedupedai.com/mcp
Every Monday: the 5 hottest repos and exactly how to use them. Free, one email a week, unsubscribe anytime.
Get them freeShare: X LinkedIn Hacker News Email Copy link
Third-party repos we don't own. AI-written use cases can be wrong. Disclaimer