# GLADIATOR (working name) - real-world tool combat results, machine-readable
# Every claim carries a date, an uncertainty interval, cost, and limits.
# Statuses: OK (replicated ranking) | LOCAL_VALIDITY (single block) | INDETERMINATE (no rank).
# Categories prefixed 'fixture-' are synthetic demo data validating the machinery, not evidence.
# Schema: schemas/result.v1.json in the repository. Query answers can be INSUFFICIENT_EVIDENCE.

## Canonical JSON (source of truth)
- https://torneo.ai/results/fixture-widgets/latest.json
- https://torneo.ai/results/fixture-widgets-smalln/latest.json
- https://torneo.ai/runs/GLAD-FIX-001/result.json
- https://torneo.ai/runs/GLAD-FIX-IND/result.json

## Pages
- https://torneo.ai/method : how results are produced (frozen protocol, intervals, replication)
- https://torneo.ai/limits : what these results can and cannot tell you
- https://torneo.ai/c/fixture-widgets
- https://torneo.ai/c/fixture-widgets-smalln
- https://torneo.ai/runs/GLAD-FIX-001
- https://torneo.ai/runs/GLAD-FIX-IND

## MCP server (streamable HTTP)
- https://torneo.ai/api/mcp : tools list_categories, get_results, get_run, explain_limits