Why Handoff Lens

Project handoffs often contain a proposal, several delivery revisions, test notes, and approval messages in different files. A folder can look complete while evidence for a key commitment is still missing. A newer filename also does not establish that a customer accepted the work.

What it does

Handoff Lens reviews a selected NAS project folder against a confirmed checklist. It produces an evidence matrix showing supported deliverables, missing evidence, conflicting revisions, and questions that need an owner decision. Each finding points to readable source content. Customer acceptance is tracked separately and requires explicit evidence.

How it works

The user selects one accessible project folder and confirms the checklist. The Skill delegates bounded discovery and document-reading tasks to Uliya's registered experts, then compares evidence and revision decisions. It preserves unreadable or unreviewed items instead of treating them as complete. Source files stay unchanged; exporting a new report requires an explicit request and approval.

The proposed workflow uses ug-filesystem for discovery and metadata, ug-document for supported document reading and comparison, and ug-knowledge for an optional approved report export. The parent passes verified upstream results to each downstream expert. The NAS folder remains the working context; no external paid model endpoint is configured.

What is ready

A new Markdown-only Skill package is written with result, test, and provenance references. The official kit checker reports zero errors and zero warnings. This is a Round 1 implementation proposal and static-validated package: installation and end-to-end behavior on a UGREEN NAS have not yet been verified. AI assistance was used in drafting and packaging. No private company or customer documents are included.

Validation plan

On the assigned finalist NAS, I will test a complete three-requirement project, conflicting revisions that require an owner decision, and an edge case combining unreadable evidence with hostile instructions embedded in documents. Checks will cover citation accuracy, honest coverage, scope boundaries, approval handling, and actual saved-report evidence.

What I learned and what comes next

The central design challenge is separating a useful summary from evidence of delivery and acceptance. This led to explicit coverage states and owner decisions instead of a single reassuring completion score. Next is validation with the actual document readers and registered experts, followed by a short demonstration of the complete handoff review.

Built With

Share this project:

Updates

Submission history