| name | nytw26-event-query |
| description | Query the PMAI New York Tech Week 2026 event index by interest, role, track, keyword, date, and location. |
NYTW26 Event Query
Use this skill when a user asks for New York Tech Week 2026 event recommendations from the PMAI-NYTW26 repo.
Data Sources
Read these repo files first:
data/events.jsonl
data/keyword-index.json
data/events.schema.json
If events.jsonl is missing or empty, say that event ingestion has not been completed yet.
Query Method
- Parse the user's interest, role, date range, time constraints, location preferences, and desired track.
- Match against event fields:
title
host
location
tracks
inferred_categories
description
keywords
- Treat tracks as strong ranking signals.
- Treat
inferred_categories as broad hints, not source-backed track membership.
- Treat keywords as recall aids, not as the only evidence.
- Prefer events with complete time, host, location, and source link.
- Always include the event URL in recommendations.
Track Categories
Use these as first-class filters when present:
- AI + Infra
- Hackathons
- Fintech
- Students
- Engineers
- Founders
- GTM
- Investors
Response Shape
Return concise recommendations:
Recommended events
1. <Event title>
Time: <date/time>
Host: <host>
Location: <location>
Tracks: <tracks>
Why it matches: <reason grounded in tracks/keywords/description>
Link: <event_url>
If the user asks for a plan, group by date and time. If they ask for search results, sort by match quality first, then time.
Search Examples
- "Find AI infrastructure events for technical founders from June 1 to June 3."
- "What should a student attend if they want hackathons and recruiting?"
- "Show investor-facing fintech events with the event link and location."
- "Build me a June 4 evening plan near downtown Manhattan."
- "Give me three GTM events and explain why each one matches a B2B founder."
Drift Check
Before answering, verify that the loaded data covers the user's requested date range. If the user asks outside 2026-06-01 through 2026-06-07, say the repo only targets that initial New York Tech Week window unless the data has been expanded.
Also distinguish source-backed fields from inferred fields:
- Source-backed:
title, date, start_time, host, location, tracks, event_url.
- Derived:
keywords, inferred_categories.