30.0 / 30
What changed in the harness
Selection accuracy 100→98, token cost up 1%, unconfirmed writes 100%→100%.
Category breakdown
Where the score comes from.
Earned points across the four signals Gradable measures. Safety and Legibility are scored out of 30; Economics and Discoverability are scored out of 20.
01Safety
02Legibility
29.4 / 30
03Economics
20.0 / 20
04Discoverability
12.2 / 20
Highest-impact fix
Estimated gain +8 pointsMake target tools discoverable on the first call
Clarify tool names, decision boundaries, and required argument schemas so an agent can choose and construct the target call without exploratory steps.
Description evidence
Defects and rewrites.
1 defect found across the exposed tool descriptions. Suggested rewrites make purpose, inputs, boundaries, and returns easier for an agent to understand.
| Tool | Defect types | Suggested rewrite |
|---|---|---|
trinv-chercher-parcelle |
params_unexplained no_return_description |
Recherche des parcelles cadastrales ayant une certaine surface au sein d'une commune française. Le paramètre `fragment` désigne la commune : fournissez un fragment de son nom (par exemple « BEURD » ou « BEDOUS », comme pour la recherche de communes) ou son code INSEE pour la cibler précisément. Le paramètre `area` est la surface recherchée, exprimée en mètres carrés. L'outil retourne la liste des parcelles cadastrales qui correspondent à la commune et à la surface indiquées. Exemples : « Y a-t-il une parcelle de surface 247 m² dans BEDOUS ? » ; « Je cherche une parcelle dans BEURD faisant 333 m² ». |
Selection evidence
Confusable tool pairs.
1 pair where similar names or overlapping descriptions may send an agent toward the wrong tool.
| Tool A | Tool B | Confidence | Why they collide |
|---|---|---|---|
trinv-chercher-commune |
trinv-chercher-parcelle |
low | The descriptions clearly separate scope: commune returns commune info (INSEE, coordinates) from a name fragment, while parcelle requires an additional mandatory `area` input and returns cadastral parcels. Both share the 'chercher + fragment' pattern and a task naming a commune could superficially seem to fit both, but the presence/absence of an area requirement and the phrase 'parcelle de surface' disambiguate in nearly all natural-language phrasings; the provided examples ('chercher BEURD' vs 'parcelle de surface 247 m² dans BEDOUS') map cleanly to one tool each. |
Compare the field