yc-company-scouting-ops

name	yc-company-scouting-ops
description	Scout YC companies, batches, and Launch YC signals for standout startups and market patterns.
memory_tags	["domain:startup-research","source:y-combinator","workflow:company-scouting","skill_role:researcher","repo_boundary:tools","risk:low"]

YC Company Scouting Ops

Trigger

Use when:

• the user asks to check YC, Y Combinator, Launch YC, or a YC batch
• the user wants to find interesting YC companies, standout wedges, or batch themes
• the user wants a Product Hunt or X-style summary of what is happening in YC
• the user wants to compare YC companies by category, batch, market, traction proxy, or idea fit

Do not use when:

• the task is general market sizing; use market-landscape-research-ops
• the task is revenue estimation from public signals; use revenue-signal-research-ops
• the task is naming or wedge selection after research is already done; use naming-ops or idea-wedge-selection-ops

Inputs

• Required: target scope, such as latest batch, a specific batch, category, company, market, or "what is interesting right now."
• Optional: filters for AI, devtools, vertical SaaS, healthcare, fintech, consumer, local services, content tools, or personal usefulness.

Workflow

1. Define the candidate universe before ranking:
   • Record the batch/category/status filters, fetch time, and official YC surfaces included.
   • For a batch scan, enumerate the complete matching YC Companies directory result set; for a launch scan, enumerate the bounded Launch YC window.
   • Do not let companies encountered on X, Hacker News, or third-party lists silently become the whole universe.
2. Fetch current data from primary YC surfaces first:
   • YC Companies directory for company, batch, industry, description, founder, status, and website data.
   • Launch YC for launch posts and public launch positioning when available.
3. Resolve canonical records:
   • Link each shortlisted company to its official YC company record and company website.
   • Cluster renamed companies, duplicate launch posts, and multiple products from the same event rather than double-counting them.
4. Add secondary context only when useful:
   • company website
   • founder/company X or LinkedIn
   • Hacker News launch/comment threads
   • YCDB, YC Trends, or other public batch-analysis sites as secondary summaries
5. Identify "doing well" cautiously. Use public proxies rather than pretending to know revenue:
   • recently launched with clear positioning
   • strong HN/Product Hunt/X discussion
   • fast hiring, credible customers, funding news, or visible usage
   • repeated category pattern across a batch
   • unusually sharp wedge or distribution advantage
6. Cluster companies by:
   • buyer and user
   • workflow/job-to-be-done
   • wedge type
   • category density in the batch
   • overlap with the user's current projects or interests
7. Run a coverage gate before the short read:
   • Confirm every company above the chosen launch/traction/category threshold was shortlisted or explicitly omitted.
   • Record omitted high-signal companies and a reason such as duplicate, off-scope, insufficient current evidence, or already covered.
   • Label directory pagination, unavailable Launch YC pages, stale company metadata, and blocked secondary sources as source gaps.
8. Produce a short read:
   • standout companies
   • batch themes
   • suspiciously crowded categories
   • ideas worth studying or adapting
   • companies to follow on X/HN/Product Hunt if public signals matter
9. If the user asks for durable tracking, save a dated note in the private repo, for example:
   • <private-repo>/captures/startups/yc/trends/YYYY-MM-DD.md

Outputs

• A ranked shortlist of interesting YC companies with source links.
• A batch/category theme summary.
• A "why this matters" read for product ideas, content, or market selection.
• Caveats around evidence quality and which signals are only proxies.
• Snapshot scope, source gaps, and a compact omitted-candidate audit.

Source Guidance

• Prefer official YC surfaces for company and batch facts.
• Treat third-party YC databases and analysis sites as helpful indexes, not canonical truth.
• Always date the scan because YC batches, launches, and company metadata change.

Boundaries

• Public specs live in liferepo.
• Private notes and saved snapshots live in <private-repo>.
• This skill only owns reusable scouting behavior.
• Keep the skill reusable: no hardcoded personal handles, emails, account ids, credentials, or user-specific defaults.