Onboarding path (recommended)
- Install
@icodenet/eval-dashboards. - Run
eval-dashboards init --preset=agent-quality --write --playbook. - Emit
eval-report/v1files into.evals_output/. - Run
lint,check, andreport. - Publish static output to GitHub Pages.
Versioned docs sections
- Getting Started: install +
initjourney for vitest/jest/node/python. - CLI: command reference and flags for daily workflows.
- Schema/Taxonomy: contract fields and completeness targets.
- CI Gates: gate recipes and baseline strategy options.
- Publishing: static publish targets and Pages workflow shape.
- Integrations: links to runnable examples and adapters.
- assistant-ui case study: external-repo adoption with reproducible eval-dashboards commands.
- Adopt now: simple path to adopt with your existing runner and CI.
- Teach labs: delivery-stage labs + FDE workflow using concrete report artifacts.
- Adoption map: candidate PR-style change sets for three existing-runner repos (openai/evals, microsoft/autogen agbench, huggingface/lighteval). Not endorsed or accepted by those projects.
- API Reference: public TypeScript exports for package consumers.
- Search: client-side docs search across v1 pages.
- Benchmark packs: versioned suite bundles for safety/tool-routing/groundedness.
- Adoption feedback loop: weekly docs-adoption signal checks.
- Docs friction backlog: prioritized documentation blockers with owners.
- SOUL.md: product intent and non-negotiable principles.