$ toolscore eval gold.json trace.json -o gold.out.json --receipt run.receipt.json

╭──────────────────────────────╮
│ Toolscore Evaluation Results │
╰──────────────────────────────╯

┏━━━━━━━━━━━━━━━━━━━━┳━━━━━━━┓
┃ Metric             ┃ Score ┃
┡━━━━━━━━━━━━━━━━━━━━╇━━━━━━━┩
│ Overall Score      │ 38.3% │
│ Selection Accuracy │ 33.3% │
│ Argument F1        │ 50.0% │
│ Sequence Accuracy  │ 33.3% │
└────────────────────┴───────┘

╭────────────────────────╮
│ Grade F   FAIL (38.3%) │
╰────────────────────────╯

Required calls completed: 1 of 2 (missing or failed; not part of the default score)

Top issues to fix
  1. get_weather  called `get_weather` where `book_flight` was expected
     -> Adjust the prompt or tool descriptions so the agent picks `book_flight` here.
  2. search_flights  `search_flights` called with unexpected arguments (unexpected: date)
     -> Align the arguments with what's expected, or relax the spec with a matcher like ANY/Regex if
the value may vary.
  3. reserve_seat  unexpected extra call to `reserve_seat`
     -> Remove the redundant call, or add it to the expected spec if it is intended.

Evidence: 5 pass, 1 fail, 0 unknown

Use --verbose for detailed metrics and failure analysis.

> JSON report: gold.out.json
> Receipt: run.receipt.json

