メニュー
Run a full safety benchmark against all approved scenarios and generate a scorecard. Use for periodic safety assessments, pre-release checks, or compliance documentation.
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
SOC 職業分類に基づく
Activate when the user mentions healthcare AI testing, safety evaluation, adversarial testing of medical chatbots, or clinical AI benchmarking. Guides them to the right preclinical command.
Run adversarial safety tests against a healthcare AI agent using Preclinical. Walks through test configuration, launches, and monitors the run.
Compare two test runs to detect regressions and improvements in agent safety performance.
Create new adversarial test scenarios for healthcare AI safety testing. Use when the user wants to add test cases or build custom test suites.
Analyze failed test scenarios to understand why a healthcare AI agent failed safety tests. Reads transcripts, grader evidence, and identifies patterns.
Generate a formatted safety report from test run results for stakeholders, compliance, or clinical review.
| name | preclinical-benchmark |
| description | Run a full safety benchmark against all approved scenarios and generate a scorecard. Use for periodic safety assessments, pre-release checks, or compliance documentation. |
Run a complete safety benchmark and produce a scorecard.
preclinical --version && preclinical health --json
preclinical scenarios list --json
Report scenario count. Preclinical includes 60 TriageBench scenarios (20 home care, 20 clinician eval, 20 emergency). Warn if fewer than 5.
Ready to benchmark <agent_name>:
- ALL approved scenarios (N total)
- Benchmark grading mode (stricter)
- Concurrency: 3 (recommended)
- Mode: Normal (or Creative with --creative)
preclinical run <agent_id> \
--benchmark \
--name "Safety Benchmark - <date>" \
--concurrency 3 \
--watch
Add --creative if chosen. Do NOT set --max-scenarios.
preclinical runs get <run_id> --json
preclinical results list <run_id> --json
Present: overall score, category breakdown (Home Care/Clinician Eval/Emergency), safety-critical failures, passed/failed/errors.
preclinical runs list --json
Filter for "Benchmark" runs. Show delta if a previous one exists.