[Bug] JSON schema validation against the Assessment submission which has less than 25 attributes fails hard in the Leaderboard CI

[kami619]

Bug Description

The JSON schema (assessment-schema.json) rejects valid assessments generated with --exclude flags because it hardcodes attributes_total as exactly 25 and findings array length as exactly 25.

To Reproduce

1. Run agentready assess . --exclude architecture_decisions --exclude openapi_specs
2. Submit the assessment to the leaderboard via PR
3. CI fails with: attributes_total: 25 was expected and findings: [...] is too short

See PR #301 for a real example.

Expected Behavior

Assessments using --exclude should either:

• Validate successfully (if exclusions are allowed for leaderboard)
• Fail with a clear message at assessment time (if exclusions are disallowed)

Actual Behavior

The assessment generates successfully, but schema validation fails at submission time with a confusing error.

Environment

• agentready: 2.28.0
• Python: 3.12
• CI: GitHub Actions (Ubuntu 24.04)

Possible Solution

Option A (minimal change): Add a leaderboard-specific check in the workflow that rejects exclusions early with a clear message, keeping the schema strict:

- name: Reject excluded assessments
  run: |
    TOTAL=$(jq -r '.attributes_total' "$ASSESSMENT_FILE")
    if [ "$TOTAL" -ne 25 ]; then
      echo "::error::Leaderboard requires all 25 attributes. Found $TOTAL (did you use --exclude?)."
      exit 1
    fi

This preserves the intent that leaderboard submissions must be comparable (all 25 attributes), while giving users a clear error instead of a schema validation failure.

Option B: Relax the schema to allow attributes_total <= 25 — simpler but allows non-comparable submissions.