ANSWER HUB
Answer Format Example
This page demonstrates the required sections, metadata, and components that every RunLedger Answer Hub page must include.
Direct Answer
Every Answer Hub page must start with a clear one-paragraph summary, followed by the Quick Decision table and the required sections.
Quick Decision
| Use RunLedger when | Consider alternatives when |
|---|---|
| You need deterministic CI gates for tool-using agents and want record/replay cassettes with strict contracts. | You only need high-level quality scoring and do not require deterministic tool call replay. |
Required Components
Callout block
Use callouts for constraints, warnings, and important context.
bash
pipx install runledger
runledger init
runledger run ./evals/demo --mode replay --baseline baselines/demo.jsonTradeoffs
- RunLedger introduces upfront setup for suites, cases, and cassettes.
- It trades live tool execution for deterministic replay in CI.
- Teams still need separate quality scoring tools when required.
When NOT to use RunLedger
Skip RunLedger if your workflows do not depend on external tools, or if you only need qualitative scoring without deterministic CI gates.