diff --git a/docs/product/roadmap.md b/docs/product/roadmap.md index d76d775..25077b7 100644 --- a/docs/product/roadmap.md +++ b/docs/product/roadmap.md @@ -13,10 +13,8 @@ - Identify which CI/review surface creates the most adoption friction. - Calibrate evaluator policies against qualified human labels. -- Calibrate fixed and sequential statistical policies on design-partner replay +- Calibrate fixed and sequential statistical policies on reviewed replay bundles; measure false-pass, false-block, cost, and stopping behavior. -- Measure repeat use, task success, time saved, disagreement, and critical - incidents in a consented design-partner pilot. - Review supported Python versions, provider adapters, and release automation for drift before the next package version. diff --git a/docs/project/status.md b/docs/project/status.md index a2e72d2..b7ad63f 100644 --- a/docs/project/status.md +++ b/docs/project/status.md @@ -2,23 +2,49 @@ ## Current state -- Stable package: `1.0.0`. -- Public GitHub milestone: `v1.0` (new baseline). +- Stable package: `1.1.0`. +- Public GitHub milestone: `v1.1`. - Repository license: MIT. -- Released runtime contract: `1.0.0`. -- Current work: statistical regression milestone accepted and published as the - `v1.0` GitHub Release and `1.0.0` Python package on 2026-07-15. +- Released runtime contract: `1.1.0`. +- Current work: the three-minute adoption milestone was accepted and published + as the `v1.1` GitHub Release and `1.1.0` Python package on 2026-07-16. ## Compatibility Existing evaluation schemas, metric meanings, and deterministic PASS/BLOCK behavior remain compatible. New replay, statistical, sequential, drift, -provenance, and baseline-manifest contracts are opt-in additions in 1.0.0. The -open-core boundary remains unchanged. +provenance, and baseline-manifest contracts remain compatible. Source-freshness +and abstention-contract evaluation are opt-in additions in 1.1.0. The open-core +boundary remains unchanged. ## Released baseline -### Acceptance evidence +### v1.1 acceptance evidence + +The 2026-07-16 acceptance lane completed on local and GitHub-hosted runners: + +- `ruff check .` and 208 tests passed with third-party pytest plugin autoload + disabled; CI passed on Python 3.11, 3.12, and 3.13. +- A clean, credential-free `uvx ragops demo` run from outside the repository + installed `1.1.0`, produced a BLOCK decision, and wrote Markdown and HTML + reports. +- The repository's reusable gate blocked the intentional regression and then + passed the restored candidate. The same pull-request comment reported metric + deltas, named block reasons, and the HTML artifact link. +- Wheel and source archive builds, clean-wheel execution, deterministic and + statistical PASS/BLOCK paths, sequential decisions, evaluator drift, + baseline integrity/signature verification, API authentication/body bounds, + secret review, and desktop/mobile visual checks passed. +- The release workflow produced checksums, a reproducible CycloneDX SBOM, and + build provenance. PyPI Trusted Publishing promoted the exact GitHub Release + distributions without rebuilding. +- GitHub Pages deployed the merge commit and exposes the one-command demo, + result screenshots, honest comparison, Failure Zoo, and concise roadmap. + +The owner accepted this evidence and authorized release `v1.1` and package +`1.1.0` on 2026-07-16. + +### v1.0 acceptance evidence The 2026-07-15 pre-acceptance lane completed on Python 3.12: @@ -73,6 +99,6 @@ Acceptance checklist for the next milestone: ## Recommendation -The current public baseline is GitHub milestone `v1.0` and Python package -`1.0.0`. Keep subsequent compatible work under `Unreleased`; create `v1.1` and -publish `1.1.0` only after a new owner-approved acceptance lane. +The current public baseline is GitHub milestone `v1.1` and Python package +`1.1.0`. Keep subsequent compatible work under `Unreleased`; publish another +version only after a new owner-approved acceptance lane.