eliziff narrows Beaver around Canadian law

The fork is making an early product bet: build for one legal setting before adding breadth.

complianceworkflow

These commits do not ship a new client-facing feature, but they set the boundaries for the work that follows. eliziff has explicitly scoped Beaver to Canadian law, deferred role-aware onboarding until the product has usable presets, and treated accessibility as a release requirement rather than a later polish job.

There is also a deliberate preference for a measured, modular build instead of premature expansion. For legal teams, that matters because jurisdiction, workflow and accessibility assumptions tend to show up everywhere once a product starts handling real work. The GitHub notes are worth a look for anyone assessing how firmly this fork intends to stay focused.

So what Canadian legal-product builders and firms evaluating local-fit tools should care: this fork is choosing a narrower lane early, with fewer implied promises outside it.

View this fork on GitHub →

Spotted something wrong? Or know the PR text has fresher detail than the writeup above?

Commits in this thread

6 commits from eliziff/Beaver, oldest first. Source extracted verbatim from the harvested git log.

SHA	Subject	Author	Date
`a35eb7f1`	docs: defer role-aware onboarding until presets are useful	Eli Ziff	2026-07-27	↗ GitHub
`fa63307b`	docs: make accessibility a cross-cutting release gate	Eli Ziff	2026-07-27	↗ GitHub
`ad443ac4`	docs: set a measured modular-monolith budget	Eli Ziff	2026-07-27	↗ GitHub
`7d81eab6`	docs: compare legal skill/plugin ecosystems to Mike	Eli Ziff	2026-07-27	↗ GitHub
commit body Survey Lawve AI, awesome-legal-skills, and anthropics/claude-for-legal against Mike's 24 built-in workflows. Identify six categories of published "skill" that are deterministic computation expressed as prose, propose three workflows that pass the Class D rigour test, and specify a versioned legal rule pack engine as the deterministic replacement for the deadline, statutory-classification, and list-screening clusters. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_0196DAeXBKv4jK7D3oFzbZFH
`47481fc4`	Add session-learned agent guidance to AGENTS.md + CLAUDE.md	Eli Ziff	2026-07-28	↗ GitHub
commit body Terse working notes future models need in this codebase: local A2AJ bulk corpus location (check before network fetches; consolidations elide amendment prose), measurement-gated grammar changes and refusal-over-guessing for the deterministic legal-text modules, LAB harness LAB_SANDBOX_ENGINE=docker + stub API key warning, Windows shell traps (grep -P locale, vitest NULs, cp1252, /c/ paths), and shared-tree git etiquette incl. the env -u GITHUB_TOKEN push fix. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018bhP89cYSdEEjHNPnxacns
`dff754b0`	Clarify focus on Canadian law in project description	eliziff	2026-07-29	↗ GitHub
Added context about Canadian law to the README.

Capture this thread into my fork

Download a single Markdown prompt that tells Claude how to port every commit above into your working tree — adapting paths and structure to match your repo. Run it via claude -p < capture-thread-827.md from inside the repo you want the changes in.

⬇ Download capture-thread-827.md