HUNT Local Scout

Inspiration

Many builders keep contest PDFs, RFPs, and application screenshots scattered across folders. HUNT Local Scout brings the contest-discovery experience directly to your AI NAS—no cloud dependency required.

What it does

HUNT Local Scout is a UGOS Skill that:

• Watches a user-selected folder (e.g., /Documents/HUNT/inbox)
• Extracts deadline, prize, and URL heuristics from PDFs and text
• Ranks opportunities by expected value (EV) on-device
• Writes a ranked opportunities.md and JSON ledger locally
• Never leaves the NAS unless the user opts into sync

How we built it

Built on the HUNT contest-agent platform, adapted for UGOS:

• Python skill scaffold for UGREEN AI NAS
• Local LLM integration for PDF parsing (no API keys)
• EV ranking algorithm (same thesis as HUNT web product)
• File-watcher pattern for automatic updates

Challenges we ran into

• UGOS Skill packaging format (Round 1 uses conceptual scaffold)
• Balancing privacy-first architecture with useful automation
• Adapting web-first HUNT UI to NAS headless/dashboard UX

Accomplishments

• Privacy-first contest discovery for NAS users
• Zero external API dependencies
• Reusable skill pattern for other document-workflow automation

What we learned

UGOS makes it possible to bring intelligent automation to personal data without compromising privacy—perfect for builders who want control over their opportunity pipeline.

What's next

• Full Skill Package ZIP with UGOS runtime
• Desktop/mobile companion app for cross-device sync
• Community skill marketplace for HUNT workflows

Built With

• llm
• local
• python
• sdk
• ugos