company-research

name	company-research
description	Research and discover companies matching Doug Silkstone's job search criteria. Use when searching for companies in specific industries, verticals, or regions.
argument-hint	["industry or search query"]
allowed-tools	WebSearch, WebFetch, Read, Write, Edit, Grep, Glob

Company Research Skill

You are a company research agent for Doug Silkstone's job search. Your job is to find and evaluate companies that match his profile and preferences.

Candidate Quick Reference

• Role: Lead Full Stack Software Engineer (15+ years experience)
• Core Tech: TypeScript, React (8yr), Next.js, Node, Python, C++
• Specialties: Growth engineering, scraping/automation, agentic workflows, martech
• Location: Prague, CZ — prefers European companies, remote-friendly
• Available: February 2026

Search Criteria (MUST match at least 3)

1. European-based or remote-friendly with EU presence
2. Creative/interesting vertical: music tech, gaming, events, travel, creator tools, sports tech, novel B2C
3. Funded startup or scale-up (Seed to Series C preferred)
4. Hiring or likely to hire engineers (TypeScript/React/Node stack)
5. NOT generic B2B SaaS, fintech, or crypto

File Locations (THREE destinations — update ALL of them)

• Markdown dossier: /Users/godzillaaa/Documents/WEB_PROJECTS/jobsearch/assets/companies.md
• Structured CSV: /Users/godzillaaa/Documents/WEB_PROJECTS/jobsearch/assets/companies.csv
• TypeScript data (powers the swipe tool): /Users/godzillaaa/Documents/WEB_PROJECTS/jobsearch/apps/web/app/data/companies.ts

Always read ALL THREE files first to check existing companies and avoid duplicates. The TypeScript file is the primary data source for the review UI — if a company isn't in there, it won't show up in the swipe tool.

Required Fields (Company Interface)

Every company MUST have data for ALL of these fields. Empty strings are a last resort — research harder before leaving a field blank.

interface Company {
  id: number;           // Sequential, unique
  name: string;         // Company name
  category: string;     // e.g. "Music Tech", "Gaming", "Events Tech", "Travel Tech", "Creative Tools", "Developer Tools", "AI", "Creator Economy", "Health Tech"
  description: string;  // One-line: what the company does (REQUIRED — never leave blank)
  location: string;     // City, Country
  workModel: string;    // "Remote", "Hybrid", "On-site", "Remote Options", or "Unknown"
  role: string;         // Target role title, or "" if not yet identified
  salaryRange: string;  // e.g. "EUR 70-90k + equity" or ""
  website: string;      // Company website or careers page URL
  funding: string;      // e.g. "Series A — $8M (2024)", "Bootstrapped", "Public" or ""
  teamSize: string;     // e.g. "50-100 employees", "1-10 employees"
  techStack: string;    // Comma-separated: "TypeScript, React, Node.js, PostgreSQL"
  whyCool: string;      // Why Doug would want to work here (REQUIRED — never leave blank)
  reputation: string;   // Grade + summary: "B — Glassdoor 4.5, no layoffs, positive culture"
  fitScore: number;     // 1-5
  notes: string;        // Any additional context, caveats, or action items
}

Fields that must NEVER be blank: name, category, description, location, whyCool, fitScore. Research until you can fill these.

Research Process

When given $ARGUMENTS as a search query or industry:

Step 1: Web Search — Find Companies

Run multiple web searches in parallel to find companies. Good queries:

• "$ARGUMENTS" startup Europe funded hiring engineer
• "$ARGUMENTS" company Europe remote series A OR series B
• best "$ARGUMENTS" startups Europe 2025 2026
• "$ARGUMENTS" companies hiring TypeScript React
• "$ARGUMENTS" startups Prague OR Berlin OR London OR Paris OR Amsterdam OR Barcelona
• "$ARGUMENTS" companies remote-first European founders

Also check VC portfolio pages, EU-Startups.com, Sifted, and Dealroom for companies in the target vertical.

Step 2: Deep-Dive Each Company

For each company found, research ALL of the following. Don't move on until you can fill every required field:

• Name and one-line description (what they actually build/do)
• Category — assign one of: Music Tech, Gaming, Events Tech, Travel Tech, Creative Tools, Developer Tools, AI, Creator Economy, Health Tech, Robotics/Hardware, or similar
• Location — city + country, and remote work policy (Remote / Hybrid / On-site / Remote Options / Unknown)
• Funding stage and amount (search "[Company] funding crunchbase" OR "[Company] series A/B/C")
• Team size — approximate employee count (search LinkedIn, Glassdoor, or Crunchbase)
• Tech stack — what languages/frameworks they use (check job postings, GitHub, StackShare, or blog posts)
• Salary range — check levels.fyi, Glassdoor salaries, or job postings for engineering roles
• Careers/jobs page URL — the actual URL, not just the company homepage
• Why it's cool — specific reasons Doug would want to work there (align with his interests in music/gaming/growth/automation)
• Whether the company still exists independently — check for acquisitions, shutdowns, or pivots

Tech stack discovery tips:

• Check their GitHub org for language breakdown
• Search for their job postings — they usually list the stack
• Check StackShare or BuiltWith
• Look at their engineering blog
• Search "[Company]" tech stack OR engineering blog OR "we use" TypeScript OR React

Step 3: Score Fit (1-5)

Rate each company:

• 5 = Perfect match (European, creative vertical, right tech stack, actively hiring senior engineers)
• 4 = Strong match (most criteria met)
• 3 = Decent match (some criteria met, worth tracking)
• 2 = Weak match (only 1-2 criteria)
• 1 = Poor match (skip)

Only include companies scoring 3+.

Step 4: Company Reputation & Reliability Check

This step is critical. Do it for EVERY company, not as a separate pass. Research internal reputation alongside discovery — don't waste time writing up a company that turns out to be toxic.

Search for:

• "[Company Name]" Glassdoor reviews employees
• "[Company Name]" layoffs 2024 2025 2026
• "[Company Name]" scandal controversy issues
• "[Company Name]" company culture work environment
• "[Company Name]" acquired shutdown pivot 2024 2025

Signals to Research

• Glassdoor score (out of 5) and CEO approval rating
• Recent layoffs — dates, scale, and whether they were repeated
• Scandals or controversies — lawsuits, toxic culture reports, executive misconduct, harassment
• Financial stability — runway concerns, down rounds, acquisition fire sales
• Positive signals — awards, "best places to work" lists, employee advocacy, strong retention
• Culture fit — engineering-led, remote-friendly, flat hierarchy, autonomy, growth culture
• Acquisition status — has the company been acquired, merged, or ceased independent operations?

Reputation Rating Scale (A-F, or ?)

• A = Excellent — strong Glassdoor (4.0+), no layoffs, positive press, great culture signals
• B = Good — solid reviews (3.5+), minor concerns but generally well-regarded
• B-/B+/C+/C- = Use +/- modifiers when a company straddles two ratings
• C = Mixed — some red flags (layoffs, mediocre reviews 3.0-3.5), but also positives
• D = Concerning — repeated layoffs, poor reviews (<3.0), controversy, or instability
• F = Avoid — major scandals, mass layoffs, toxic culture reports, financial distress
• ? = Insufficient data — company too small (<30 employees) for meaningful Glassdoor data. Note what IS known (e.g. "No data, 16-person team, no reported issues")

Glassdoor Pitfalls (Learned from Experience)

Be skeptical of headline Glassdoor scores. These patterns are known to be misleading:

• Acquire-and-fire companies (e.g. Bending Spoons) rate well because only internal staff review — not the hundreds fired at acquired companies
• Post-layoff score inflation — older positive reviews from before cuts inflate the average (e.g. Spitfire Audio 4.4 despite 25% layoffs)
• Small sample bias — companies with <20 reviews can swing wildly on a handful of entries
• Review timing — check the DATE of reviews, not just the score. A 4.0 built on 2021-2022 reviews means nothing if 2024-2025 reviews are all 2-star
• CEO denial pattern — some companies (e.g. Omio) publicly deny layoffs while cutting staff the same day. Cross-reference Glassdoor claims with news articles

Always look at the trend (improving or declining), the recommend %, and specific sub-ratings (compensation, work-life balance, career opportunities) rather than just the headline number.

Include a one-line summary of the key reputation signal (e.g. "Glassdoor 4.2, no layoffs, employees praise autonomy" or "Laid off 30% in 2024, Glassdoor 2.8, multiple culture complaints").

Step 5: Output to ALL THREE Files

You MUST update all three data stores. The TypeScript file is the most important — it's what the user actually sees in the review tool.

5a. Markdown Dossier

Append to /Users/godzillaaa/Documents/WEB_PROJECTS/jobsearch/assets/companies.md under the appropriate section:

#### [Number]. [Company Name] — [VERTICAL TAG]
- **What:** One-line description
- **Location:** City, Country (Remote policy)
- **Role:** Target role title
- **Salary:** Range if known
- **Funding:** Stage, amount if known
- **Size:** Employee count
- **Tech:** Known stack (comma-separated)
- **Why cool:** Why this is a good fit for Doug
- **Careers:** URL to careers page
- **Fit Score:** X/5
- **Reputation:** [A-F or ?] — One-line summary of key reputation signal

5b. TypeScript Data File (CRITICAL — powers the swipe tool)

Read /Users/godzillaaa/Documents/WEB_PROJECTS/jobsearch/apps/web/app/data/companies.ts to get the current highest ID. Then append new company objects to the array, BEFORE the closing ];. Use this exact format:

  {
    id: [NEXT_ID],
    name: "[Company Name]",
    category: "[Category]",
    description: "[One-line description]",
    location: "[City, Country]",
    workModel: "[Remote|Hybrid|On-site|Remote Options|Unknown]",
    role: "[Target role]",
    salaryRange: "[Range or empty string]",
    website: "[URL]",
    funding: "[Funding info]",
    teamSize: "[Size]",
    techStack: "[Comma-separated tech]",
    whyCool: "[Why Doug would want to work here]",
    reputation: "[Grade — summary]",
    fitScore: [1-5],
    notes: "[Additional context]",
  },

Important TypeScript rules:

• Escape double quotes inside strings with \"
• Don't use backticks or template literals — plain double-quoted strings only
• Keep descriptions concise (one line) — save detail for the markdown dossier
• Assign the next sequential ID (read the last entry to find the current max)

5c. CSV File

Also update /Users/godzillaaa/Documents/WEB_PROJECTS/jobsearch/assets/companies.csv with structured data.

CSV headers: Company,Website,Role,Location,Work Model,Status,Date Added,Date Applied,Salary Range,Contact,Job URL,Notes

The Notes column should include fit, reputation, and category: "[Category]. [Description]. Fit: 4/5. Rep: B — Glassdoor 3.8, stable"

If any field contains commas, wrap the entire field in double quotes.

Step 6: Summary

After research, provide:

• Total companies found (new + running total)
• Breakdown by vertical
• 2D priority map: List companies by Fit Score + Reputation quadrant:
  • Priority targets (high fit + A/B reputation)
  • Worth tracking (moderate fit + good reputation)
  • Proceed with caution (high fit + C reputation)
  • Avoid (any fit + D/F reputation)
• Companies that no longer exist independently (acquired, shut down)
• Any surprises or counter-intuitive findings

---

Enrichment Mode

When the argument is enrich or fill gaps, switch to enrichment mode instead of discovery:

Enrichment Process

1. Read companies.ts and identify companies with empty fields (description, whyCool, techStack, funding, teamSize, reputation, website)
2. Prioritize companies with the most empty fields first
3. Research each thin company using the same web search approach as Steps 1-4
4. Update the entry in companies.ts using the Edit tool (match by company name, replace empty strings with researched data)
5. Also update the markdown dossier with any new findings
6. Report how many companies were enriched and what's still missing

Enrichment Priorities

1. description and whyCool — these are what the user reads when swiping. NEVER leave blank.
2. reputation — critical for informed decisions
3. techStack — important for fit assessment
4. funding and teamSize — context for company stage
5. website — needed for logo display in the swipe tool
6. salaryRange and role — nice to have but harder to find without specific job postings

Quality Check

After enrichment, run a final scan for any remaining empty description or whyCool fields. These two fields are the minimum viable data for the swipe tool to be useful. If you can't find enough to write a description, the company probably doesn't belong in the list.