Model Routing Readiness Pack v1.0 Release Notes
What it does
Model Routing Readiness Pack turns redacted routing notes into a deterministic model routing readiness report. It helps operators verify primary and fallback models, quality thresholds, latency and cost budgets, context limits, retries, health checks, fallback tests, and rollback plans before routing policy changes. It is advisory and read-only, so the output supports review and release workflows without acting as an approval authority.
It is built for marketplace buyers who need a quick, repeatable way to inspect redacted evidence without changing runtime, configuration, billing, deployment, memory, or finance-system state.
Use cases
- AI operations teams reviewing routing changes before release.
- Cost and reliability reviewers who need budget, fallback, and stop-condition evidence.
- Platform owners who want model route readiness scored without changing runtime state.
Requirements
- Python 3 with standard-library support.
- OpenClaw workspace with the skill installed.
- Redacted local notes supplied via
--textor--notes. - No API keys required; the analyzer is local and read-only.
- Recommended OpenClaw version: current stable release.
Example usage
python3 scripts/model_routing_readiness_pack.py --text "redacted evidence with owner, validation, rollback plan, health check, and review timestamp" --json
Expected output: JSON with a verdict such as PASS, REVIEW, BLOCKED, or ERROR, a numeric evidence score, findings for missing or risky signals, and mutated_files: []. Strong evidence commonly covers route identifier, task class, primary model, fallback model, quality threshold, latency budget, cost budget, context limit, retry ceiling, health check, fallback test, and rollback plan.
Price
$4.99