Adds docstrings to open API spec (with W&B MCP + llms-full.txt)

[ngrayluna]

What

How content was generated: Uses W&B + Weave MCP, Mintlify Weave Docs MCP, llms-full.txt and Claude Code.

Claude Code summary:

I've enhanced the OpenAPI specification for the Weave Tracing API by adding:

1. API Info Section

• Updated title from "FastAPI" to "Weave Tracing API"
• Added comprehensive description explaining it's an observability API for AI/ML applications with MCP integration

2. Endpoint Descriptions

Added detailed descriptions to all 28 endpoints across 8 categories:

Service (2 endpoints):

• GET /server_info - Server information and configuration
• GET /health - Health check endpoint

Calls (8 endpoints):

• POST /call/start - Start tracing calls for MCP operations
• POST /call/end - End tracing calls with execution data
• POST /call/upsert_batch - Batch creation of call traces
• POST /calls/delete - Delete call records
• POST /call/update - Update existing call information
• POST /call/read - Retrieve detailed call data
• POST /calls/query_stats - Query call statistics
• POST /calls/stream_query - Stream call data in JSONL format

Objects (4 endpoints):

• POST /obj/create - Create objects for ML workflows
• POST /obj/read - Retrieve object data and metadata
• POST /objs/query - Query and search objects
• POST /obj/delete - Delete objects with integrity

Tables (5 endpoints):

• POST /table/create - Create structured data tables
• POST /table/update - Update table schema and data
• POST /table/query - Query table data with filtering
• POST /table/query_stream - Stream table data efficiently
• POST /table/query_stats - Get table statistics

Files (2 endpoints):

• POST /file/create - Upload files and artifacts
• POST /file/content - Download file content

Costs (3 endpoints):

• POST /cost/create - Record cost and usage metrics
• POST /cost/query - Query cost data for analysis
• POST /cost/purge - Delete cost records

Feedback (4 endpoints):

• POST /feedback/create - Create feedback records (already had description)
• POST /feedback/query - Query feedback data (already had description)
• POST /feedback/purge - Delete feedback records (already had description)
• POST /feedback/replace - Replace existing feedback

References (1 endpoint):

• POST /refs/read_batch - Batch read object references

All descriptions now clearly explain each endpoint's purpose, use cases, and how they fit into the broader Weave tracing ecosystem for AI/ML observability.