AI Workflows
Picora is built as a resource hosting platform for AI workflows. This section explains how to plug Picora into AI tools so your assistant can upload, list and manage images / videos / audio / Markdown directly from a prompt.
Why AI workflows?
Modern AI assistants (Claude Desktop, Cursor, Moraya, …) speak the Model Context Protocol — a standard way for them to call external tools. Picora exposes 12 MCP tools covering the full resource lifecycle:
- Upload images / videos / audio / Markdown documents
- List and search resources
- Get usage and quota
- Delete (with mandatory dry-run protection)
Once connected, you can ask:
“Upload the README I’m editing to Picora and give me the link.” “List all images I uploaded this week, sort by size.” “Clean up videos older than 90 days. Dry-run first.”
What’s inside
Get connected
- Overview — what is MCP, how Picora exposes tools, security model
- Claude Desktop — both stdio (local install) and HTTP OAuth (zero-install) modes
- Cursor — MCP client setup
- Moraya — one-click import + MCP cooperation
Use it well
- Prompt templates — 10 battle-tested prompts for common workflows
- Tool catalog — full reference of all 12 MCP tools (inputs, outputs, errors)
Two integration modes
| Mode | Best for | Setup |
|---|---|---|
| stdio MCP | Power users, local privacy, offline | npm install -g @picora/mcp then add to Claude config |
| HTTP OAuth MCP | Most users, zero install, multi-device | Add https://mcp.picora.me to Claude config and click “Authorize” |
Start with HTTP OAuth unless you have a specific reason to run locally.