Instadash is the hosted data surface for AI agents.
The MCP server lets any MCP-capable client (Claude Desktop, Cursor,
Windsurf, LangGraph, CrewAI, PydanticAI) publish rows, read row
slices, inspect schemas, list versions, search public grids, write
individual cells, copy/delete grids, roll back versions, export to
CSV/JSON, and pull usage stats — no database or frontend code.
Tools exposed:
• instadash_push, instadash_read, instadash_list,
instadash_get_schema, instadash_list_versions, instadash_edits,
instadash_copy, instadash_delete, instadash_update,
instadash_write_cell, instadash_rollback, instadash_export,
instadash_stats, instadash_usage, instadash_embed_snippet,
instamesh_search
The killer pattern is human-in-the-loop:
1. Agent scrapes 100 leads → publishes to Instadash
2. Human filters/edits in the browser
3. Agent reads the approved subset back via MCP
4. Agent only acts on rows the human signed off
Pricing:
- Free — 3 grids, 1k rows/grid, 7-day version history
- Builder $19/mo — unlimited grids, 100k rows/grid, API endpoints
- Agent $79/mo — 5M rows/grid, hosted MCP, webhooks, scheduled refresh
- Platform custom — BYO Cloudflare, SSO+audit logs, on-prem Worker, SLA
Built on Cloudflare.