Inspiration
Households and small businesses often need to document damage, loss, disputed deliveries, warranty issues, or insurance facts. Their receipts, warranties, photos, and camera events are fragmented across private storage, making a reliable chronology slow and error-prone.
What it does
Claim-Ready Evidence builds a private, source-linked incident report from authorized NAS documents, photos, and optional camera events. It separates confirmed facts, user statements, conflicts, and missing evidence instead of producing an unsupported narrative.
How it works
The user supplies an incident summary, time window, authorized folders, optional album or camera scope, and an output folder. The Skill gathers read-only evidence, fingerprints selected files, extracts document fields, inspects visible photo details, and optionally collects time-bounded camera evidence. It then previews a cited Markdown report and saves a versioned copy only after approval. It never moves, deletes, shares, or overwrites source files.
UGREEN AI NAS capabilities
The workflow is designed around registered UGREEN expert capabilities for surveillance timelines and evidence, filesystem search/stat/SHA-256, photo search and OCR/VLM, document extraction and reconciliation, and knowledge-base text saving.
What is implemented
A complete SKILL.md, evidence model, and validation cases are packaged. The official UGREEN Hackathon Starter Kit v0.7.3 controlled checker reports static-pass with 0 errors and 0 warnings. Device runtime validation remains pending for Round 2 because Round 1 does not require NAS access.
Privacy and safety
Evidence stays on the NAS. The Skill labels source class, chronology, hashes, conflicts, unreadable sources, and coverage gaps. It does not infer identity, intent, causation, liability, or claim value, contact third parties, or submit claims.
Development disclosure
An OpenAI language model accessed through OpenClaw assisted requirements analysis, drafting, and static review. It is not a runtime dependency and receives no NAS data through this Skill. The package contains original Markdown only and bundles no third-party code, model, data, credential, executable, or external-service dependency.
Limitations
Results depend on authorized source coverage, OCR quality, timestamps, and registered capability availability. Final save requires workflow approval, and existing reports are never overwritten automatically.
Built With
- local-ai
- markdown
- openclaw
- sha-256
- ugos
- ugreen-ai-nas
Log in or sign up for Devpost to join the conversation.