Test harness and multi-agent architecture
Three agents participate in every bootstrap test run:
| Agent | Role | Model |
|---|---|---|
| Bootstrap | The thing under test — runs the bootstrap skill | Varies (test matrix dimension) |
| Domain owner | Simulates a human answering ontology questions | Fixed (Haiku — cheap, credible) |
| Judge | Reads repo state and scores against rubric | Fixed (Opus — stable scorer) |
The domain owner is intentionally constrained: seed concept hints are anchors, not definitions.
Scenario schema
Scenarios drive test runs:
id: <kebab-case>
domain: <one-line domain description>
central_question: <the question this repo exists to investigate>
seed_concepts:
- name: <string>
hint: <one-sentence anchor>
good_bootstrap_looks_like: <1–2 sentences describing a successful result>
Snapshot path
tests/results/<id>/<model>/<YYYY-MM-DD>/
structural.json
quality.json
snapshot.json