eliziff/Beaver
Beaver is a local-first Canadian legal workspace for research, drafting, document review, and evidence-backed AI assistance.
eliziff's Beaver turns Mike into a broader legal-practice workspace, with local legal sources, matter-aware chat, document work, and a dedicated authorities area. It is explicitly positioned for Canadian legal research, including Canadian open legal data and a home-grown citator.
The fork is also pushing hard on reliability: assistants are given bounded context, evidence is treated as something to preserve and retrieve precisely, and research results are being carried into the product. Beaver is a deliberate rebrand from Mike-Canada, with local and anonymous setup paths alongside several model-backend options.
What's in it
- Canadian legal research Local, indexed Canadian legal sources support search and pinpoint retrieval without relying on live lookups.
- Citator workspace A project-scoped case graph supports note-up and version-comparison work alongside legal research.
- Matter-aware assistant Chat is organised around matters, attached documents, saved history, and carefully bounded context.
- Document editing and redlines The document workflow focuses on structured DOCX work, tracked changes, stable section references, and reliable editing.
- Book of authorities Projects can maintain a dedicated authorities workspace with saved and marked sources.
- Evidence-led AI experiments The product is testing grounded retrieval, claim checking, legal evaluation campaigns, and durable evidence handoffs.
- Local-first setup Beaver includes anonymous local setup and resilience work aimed at making the workspace usable without a cloud account.
Direction
searchdraftingknowledge-management
Activity
Threads of work (detailed view)
eliziff is making Mike's front end earn its keep
This is a concentrated effort to make a legal-AI workspace feel lighter, steadier and less repetitive without changing its core purpose.
eliziff puts a hard ceiling on legal chat's memory
This Beaver fork treats a long-running matter as a managed workspace, not an ever-growing prompt.
eliziff makes the book of authorities a real project workspace
This fork turns authority handling from a one-off task into a smoother, project-specific working area.
eliziff turns Beaver into a legal-AI testing lab
The fork is building the evidence trail needed to compare legal-AI setups without quietly mixing the results.
eliziff ships grounded legal research, then scrambles to steady it
This fork moves a research-heavy legal AI stack into the product, with same-day repairs showing where the transition got rough.
eliziff gives DOCX editing a more sceptical foundation
The Beaver fork is building a document engine that treats messy legal files as evidence to test, not structure to assume.
eliziff makes Codex a serious subscription-backed assistant
This fork turns OpenAI Codex into a persistent, recoverable assistant channel built for real working sessions.
eliziff puts legal AI on a tighter leash
This fork is testing how an AI legal assistant can ask for less context, stop digging sooner, and leave a clearer trail when it cannot finish the job.
eliziff turns Claude's command-line tool into a faster AI engine
This fork makes it possible to run legal-AI conversations through Claude's command-line interface, with session reuse cutting measured response loops by more than half.
eliziff gives legal-text detection an audit trail
This fork treats citation and legislative-text detection as something to prove against a corpus, not merely demonstrate in a happy-path demo.
eliziff turns legal AI retrieval into a reliability discipline
This fork is less interested in declaring a better answer engine than proving when its evidence can be trusted.
eliziff turns case citations into auditable legal intelligence
This fork is building a Canadian citator that aims to show not just which cases connect, but why they matter.
eliziff brings Canadian and US legal sources closer to home
This fork shifts legal research toward local, indexed source material and spends real effort keeping retrieved authorities intact.
eliziff gives Beaver one address book for its legal workflows
A small internal change makes links from assistant-guided work more likely to land users in the right place.
eliziff gives PDF evidence a fixed address
This fork is making it easier to return to the same supporting passage instead of asking the system to find it again.
eliziff narrows Beaver around Canadian law
The fork is making an early product bet: build for one legal setting before adding breadth.
eliziff rebuilds Beaver's front door for a lighter legal-AI workspace
After rejecting an initial attempt, eliziff restarted the interface overhaul the same day with a more focused compact shell.
eliziff hardens court search and PDF handling
A small security pass closes two routes by which trusted legal-research workflows could be turned against their users.
eliziff turns Beaver into a local-first legal workspace
The fork makes it possible to start and use the workspace locally without routing users through a cloud-account setup.