Web MCP tools reference
Web MCP gives your AI four tools. You don't need to call them yourself; the AI uses them when you ask in plain words. This list helps when you set up an AI agent or read a tool result.
Tools
- operations_list: Lists what the LottieFiles API can do. Can filter by type, category, keyword, or risk level. Changes your data: No.
- schema_search: Searches the API's operations, types, and fields by keyword. Changes your data: No.
- schema_details: Shows the full details of one operation or type, so the request is right the first time. Changes your data: No.
- graphql_execute: Runs the request against your LottieFiles account. Changes your data: Yes, for changes (mutations).
The first three tools only read a saved copy of the API's structure. They never touch your data. Only graphql_execute reaches your account.
How the AI uses them
The AI usually works in this order: operations_list to survey, schema_search to narrow down, schema_details to get it exact, then graphql_execute to run it.
Risk levels
Each operation has a risk level. Ask the AI to check it before running a change.
- read: Only reads data
- write: Creates or updates data
- destructive: Deletes or removes data
- billing: Affects billing or your plan
- admin: Affects workspace or account settings
Built-in prompts
Some AI apps show these as ready-made prompts:
- discover_operation: Find the right operation for a goal
- draft_graphql_query: Draft a correct query for a goal
- review_graphql_mutation: Review a change before running it
- inventory_workspace: Summarize a workspace's contents
Limits
- Requests: 60 per second
- Query depth: 10 levels
- Response size: 1 MB
- Time per request: 30 seconds
- Subscriptions (live updates): Not supported