diff --git a/docs/product-assessment.md b/docs/product-assessment.md index 9ecfed4..72f0cd9 100644 --- a/docs/product-assessment.md +++ b/docs/product-assessment.md @@ -1,8 +1,8 @@ # Base Product Assessment Status: maintained product review artifact -Last reviewed: 2026-06-17 -Base era reviewed: 1.0.x +Last reviewed: 2026-06-21 +Base era reviewed: 1.1.0 This document records a candid assessment of Base as a product and engineering effort. It is not marketing copy, and it should not drift into aspiration. When @@ -189,6 +189,65 @@ modern uv-managed Python projects can opt into `python.manager: uv`, which lets uv own the repo-local `.venv` while Base keeps discovery, activation, setup, check, doctor, and command orchestration. +### 2026-06-21 / 1.1.0 Product Review Delta + +The 1.1.0 release materially improves the product story since the 2026-06-17 +review. The strongest shipped signal is team onboarding: workspace manifests, +`basectl workspace clone --manifest `, workspace pull/configure flows, +and Project intake repair make a peer-repo workspace easier to materialize and +keep aligned. That does not make Base a team product by itself, but it gives the +"one coherent workspace, many repositories" thesis a concrete team-shaped +workflow rather than only a single-developer convenience. + +Other shipped improvements reduce earlier product objections: + +- `python.manager: uv` lets uv-managed Python repositories keep a repo-local + `.venv` while Base remains the discovery, activation, check/doctor, and + command orchestration layer. The stale Base-venv cleanup/docs work is already + tracked through the uv follow-up line (#896, #912, #931). +- `bootstrap.sh`, bottled Homebrew upgrades, and the 1.1.0 release train make + first-mile install and update behavior easier to explain than the earlier + source-checkout-only story. +- `.ai-context/` plus `basectl export-context` gives Base a repo-visible AI + context surface without assuming provider-specific upload APIs; provider + upload adapters remain a separate follow-up in #570. +- Standalone `base-bash-libs` is now a reusable primitive rather than hidden + Base internals. That strengthens the "Base is a control plane, not a pile of + private shell snippets" story, while Homebrew/core packaging remains tracked + separately in #909. + +AGPL-3.0-or-later is now an explicit adoption tradeoff. It improves the +project's open-source reciprocity stance and is coherent for a control-plane +tool, but it may slow adoption in companies with strict license review. This is +a positioning and sales-friction risk, not a reason to reopen the license choice +inside this assessment. + +Remaining risks should stay issue-backed rather than becoming a parallel +backlog here: + +- Linux remains the largest platform expansion unlock; keep the first supported + Linux runtime target in #562. +- Dev Container and Nix bridges should remain export/bridge work from Base + manifests, not a replacement for those ecosystems; this is tracked in #876. +- A local dashboard is still plausible, but should follow durable observability + and history data rather than lead it; keep the dashboard in #875 and command + history in #926. +- AI provider upload is useful only after local export stays deterministic and + privacy boundaries are explicit; keep that in #570. +- `base-bash-libs` packaging/readiness should stay in the standalone library + lane, with Homebrew core preparation tracked in #909. + +Maintainability is the refreshed watchlist. The current file-size pressure is +not in the reusable Bash standard library anymore; #873 documents why +single-file shell-library standards remain intentional. The more important +ownership pressure is in command orchestration files: `repo.sh` is roughly +3,400 lines, `setup_common.sh` is roughly 2,170 lines, `repo.bats` is roughly +1,730 lines, and `gh.bats`, `base_projects/engine.py`, and +`base_github_projects/engine.py` are all around the 1,000-line mark. The right +response is not an abstract split-everything campaign. Reduce ownership where a +stable subdomain is visible, keep tests near behavior, and use #929 for the +known `setup_common.sh` ownership-reduction path. + ## 4. Creator And Engineering Skill Assessment Assessment: at least Staff-level; plausibly upper Staff or early Senior @@ -260,6 +319,10 @@ the system without needing the creator in the loop. ## Assessment History +- 2026-06-21: Updated for the 1.1.0 review delta, including workspace clone + onboarding, uv-managed project support, bootstrap/Homebrew release maturity, + AI context exports, standalone `base-bash-libs`, AGPL adoption tradeoffs, and + the current maintainability watchlist. - 2026-06-17: Added latest product-review delta and linked follow-up issues for Linux runtime support, workspace manifest sync, artifact adapter design, and manifest command linting.