메뉴
Company and person research agent. Deep research on any target. Use /scrape [company or person name] or trigger when candidate asks to research a company or find who to contact.
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
| name | scraping |
| description | Company and person research agent. Deep research on any target. Use /scrape [company or person name] or trigger when candidate asks to research a company or find who to contact. |
Two modes: company research and person research. Always research before writing outreach. Never send a message without reading this first.
| Source | What to extract |
|---|---|
[company].com/about | Mission, founding story, team size, leadership names |
[company].com/blog | Recent product updates, tech stack mentions, culture signals |
[company].com/careers | Open roles, team names, required skills, seniority signals |
| Crunchbase | Funding stage, investors, headcount, founding year, recent rounds |
GitHub (github.com/[company]) | Languages used, repo activity, open source work, team size signals |
| LinkedIn company page | Employee count, recent posts, job listings, who works there |
Twitter/X (@[company]) | Recent announcements, tone, community activity |
| Press releases | Funding, partnerships, new products, expansions |
| ATS listing pages | Full job description, team name, required vs preferred skills |
| Source | What to extract |
|---|---|
| LinkedIn profile | Current role, previous companies, education, recent activity |
| Twitter/X | Recent posts, tone, what they care about, community engagement |
| Published papers or talks | Research areas, technical depth, public positions |
| Personal website or blog | Personality, writing style, areas of interest |
## [Company Name] — Research Brief
What they do:
[2-3 sentences: product, client type, industry, scale]
Recent news / projects:
- [Most recent significant thing — funding, product, announcement]
- [Second item if relevant]
Tech stack:
[Languages, frameworks, tools — extracted from job listings, GitHub, blog]
Culture signals:
[How they describe themselves, what they value, team size vibe]
People to contact:
- [Name], [Title] — [LinkedIn URL] — [Why relevant to candidate]
- [Name], [Title] — [LinkedIn URL] — [Why relevant]
Current/past interns found:
- [Name] — [LinkedIn URL] — [Year]
Connection points with candidate:
- [Specific overlap between candidate's background and company's work]
- [Shared context: KAUST, AI field, Saudi ecosystem, university, etc.]
Open listings: [Yes/No] — [Link if yes]
ATS platform: [Greenhouse / Lever / Workday / Ashby / Custom]
How to apply: [Direct link, recruiter name, referral path]
## [Person Name] — Profile Brief
Role: [Title at Company]
Background: [Where they studied, what they did before]
What they work on: [Day-to-day focus based on posts and activity]
Recent activity: [Most recent relevant post or announcement — last 30 days if possible]
Shared context with candidate: [Any overlap — program, university, domain, Saudi ecosystem]
Recommended message tone: [Formal / Technical peer / Community / Casual]
Recommended hook: [The one specific thing to reference in the first line]
Glassdoor exists but has very limited coverage of Saudi and GCC companies.
Use Glassdoor for:
Do NOT rely on Glassdoor for:
For Saudi-specific culture signals, better sources are:
# Research a company
python scripts/job_search.py --research "SDAIA"
# Parse a LinkedIn profile (paste the profile text, Claude reads it)
# No automated LinkedIn scraping — paste the profile text into the session
SOC 직업 분류 기준
ATS application agent. Identifies the ATS platform from a URL, generates a complete answer sheet for every form field, and optionally auto-fills the form using Playwright (Claude Code only). Run /ats [url].
CV analysis, Work History Document building, CV tailoring per role, LaTeX CV generation. Use whenever the candidate needs CV work. Read references/fit_scoring.md before tailoring.
Phase 0-1. Greet the candidate, explain the process, collect CV and initial materials. Run this first in every new session.
Cold messages and cover letters. Always research the company and person first. Ask 5 clarifying questions before writing anything. Load the company research brief and candidate profile before generating any output.
Deep adaptive discovery questioning. Builds the candidate profile and Work History Document. Run after onboarding. Ask 20-100+ questions. Do not stop until the profile is complete.
Fit scoring and opportunity comparison. Score any job listing against the candidate profile. Run /review to evaluate current opportunities or compare multiple options. Read references/fit_scoring.md.
