This example shows how changed-source mapping and advisory test-intelligence commands help an agent pick focused tests before running broader profiles.
From this directory, with Agent Maintainer installed:
git init
git add .
git commit -m "example baseline"
agent-maintainer test-intel changed --base-ref HEAD
agent-maintainer test-intel hypothesis-candidates --changed --base-ref HEAD
agent-maintainer test-intel crosshair-candidates --format json
agent-maintainer verify --profile precommit --base-ref HEADsrc/scoring/rubric.py does not clamp scores above the maximum. The repair flow
should identify tests/test_rubric.py as the focused test target, fix the
clamping behavior, and then rerun the verifier.
See expected-output.md and repair-path.md.