Why this project
What interests me here is the reliability boundary between a local draft and shared state: keeping work safe during revision conflicts and testing workspace access. This is a private pilot; live collaboration tests are still pending.
The problem
Shared documents carry two different risks: overwriting someone else’s newer revision, and exposing work outside its intended workspace. A local draft can be perfectly valid and still be unsafe to save over a document that has changed.
The approach
The workspace separates browser drafts from shared saves. A revision check decides whether a save is still current. Access boundaries are enforced independently, so knowing a document’s ID doesn’t grant access to it.
Key decisions
Reject stale saves, keep the draft
A conflicting revision produces an explicit conflict and keeps the unsaved work. Silent last-write-wins behavior feels convenient until someone loses a change.
Separate local from shared state
Editing locally and saving for everyone are different actions. Making that boundary clear tells users whether a change exists only in their browser or is visible to collaborators.
Enforce access in the database
Workspace access and immutable revisions are covered by database-level tests. Hiding things in the interface isn’t enough to protect shared data.
Evidence & limits
A stale shared revision
A save based on an outdated revision is rejected, while the editor keeps the unsaved draft.
Revision and access-control tests are recorded; a live two-browser collaboration test is still pending.
The reviewed implementation includes input validation, revision checks, and tests for workspace access, invitations, immutable revisions and stale saves. Project records include passing application and database tests; they were not rerun for this case study.
What this doesn’t prove
Live invitation acceptance and a two-browser conflict test are still pending. The hosted version of the current collaboration code hasn’t been verified yet. Live AI generation is deliberately switched off in production. Source code is private.
Evidence last reviewed 7 October 2026. This is a high-level case study: results taken from project records are kept separate from checks rerun for this review, and Chisel's code and internal reports are not published.
Takeaway
Good collaboration software protects both the shared document and the work someone hasn’t saved yet. A conflict is a state to explain and recover from, not a reason to lose data.