Scale fit across each agent's run, and lower the starter prefer weights

With fit as (cos + 1) / 2 the real run spanned about 0.55 to 0.79, so a
prefer gate at weight 0.2 moved a score more than the whole spread of fit.
fit is now the cosine min-max scaled across the roles an agent scores in
that run: the closest is 1, the farthest 0, a single role 1, and a role a
require gate rules out stays out of the range. The starter agents' prefer
weights drop to 0.05, enough to break a tie. A regression test pins the
case: two close cosines that the old mapping let a 0.05 gate reorder.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
This commit is contained in:
WolfandClaude Opus 5.5 committed 2026-09-23 12:51:08 +02:00
1 parent 59a301c351
commit 8b246c3035
5 files changed
+132 -31

No files matched your search

+2 -2
View File
@@ -30,7 +30,7 @@ agents:
ownership of production systems in a product company.
gates:
- {field: place, values: [Berlin], mode: require}
- {field: remote, values: [hybrid, remote], mode: prefer, weight: 0.2}
- {field: remote, values: [hybrid, remote], mode: prefer, weight: 0.05}
- name: Remote EU stretch
intent: >-
Staff or lead engineer role, fully remote from Europe. Platform and
@@ -38,4 +38,4 @@ agents:
leadership across teams.
gates:
- {field: remote, values: [remote], mode: require}
- {field: country, values: [DE, NL, AT, CH], mode: prefer, weight: 0.2}
- {field: country, values: [DE, NL, AT, CH], mode: prefer, weight: 0.05}