How to write, run and debug evals with reeval, the eval runner for coding agents (npm package reeval; CLI `reeval`, alias `re`). Use whenever a task mentions reeval, reeval.config.ts, *.eval.ts files, `reeval run`, evals for a coding agent, comparing skills,…
End-to-end recipe for adding a new agent harness (a CLI reeval drives in the sandbox) to reeval: the catalog entry, the adapter package, the image and pin, the contract row, the telemetry declarations, replay captures and the generated harness matrix, plus…
How to bump a harness's pinned agent CLI or AI SDK adapter in reeval: the drift scripts, the Claude Code pnpm patch and its hash in the image tag, refreshing replay captures and formats, the real contract rows and the telemetry declarations. Use when…
Read this first, before the reeval skill, whenever you're asked to set up, install, add, try or get started with reeval (the eval runner for coding agents), in a project or an empty directory, including "set up reeval here" and "add evals". It detects what's…
How to greet.
Use when formatting.
Use when answering from the guide.
The ground-truth method for checking what reeval records about a harness against what actually happened: run a real attempt, collect the gateway traffic, the CLI's on-disk state and its native stream, diff all three against the run record and transcript…