career-scout

Files (4)

• SKILL.md
• interview-frameworks.md · references/
• known-companies-israel.md · references/
• scoring-methodology.md · references/

SKILL.md (first lines)

# Career Scout

A career intelligence skill for job seekers of any profession and any seniority level.
It reads the candidate's CV to understand what role they are targeting, then ranks
companies for fit and builds tailored interview preparation guides.

The skill works for anyone: a junior frontend developer, a product manager, a UX designer,
a finance analyst, a sales lead, a VP of Marketing, a data scientist, a customer success
manager — the position is always derived from the candidate's own CV and stated goals,
never assumed.

The skill has two core workflows — **Company Ranking** and **Interview Preparation** —
preceded by a one-time **Profile Setup**.

---

## Mode 1 — Profile Setup (first time only)

Triggered when: user uploads CV/resume, shares LinkedIn URL, or says "help me find a job".

### Step 1A — Read the CV or LinkedIn

**If CV/resume uploaded (PDF or DOCX):**
Read it using the appropriate file tool. Extract silently:
- Full name, current title, current company
- **Profession / function** (e.g. software engineering, product, design, marketing,
  sales, data, finance, operations, HR, legal, etc.)
- **Seniority level** (intern, junior, mid, senior, lead, manager, director, VP, C-level)
- Years of experience and career trajectory
- Skills, tools, and domain expertise relevant to their field
  (for an engineer: languages/frameworks/cloud; for a marketer: channels/tools/campaigns;
  for a designer: tools/methods/portfolio; adapt to the profession)
- Team / org size managed, if any
- Key achievements with numbers (%, $, x, scale)
- Education and notable background
- Location

**If LinkedIn URL shared:**
Web-fetch the profile page. Extract the same fields above from the page content.
If the page is blocked, ask the user to paste their LinkedIn summary or upload their CV instead.

**The target position is inferred from the CV** — current title and trajectory point to
the natural next role(s). Confirm it with the user in Step 1B rather than assuming.

### Step 1B — Ask Targeted Questions

After reading the profile, ask 3-4 focused questions using the ask_user_input_v0 widget.
Keep it conversational — one widget call with all questions together.

**Critical rule — "Other" option:**
Every single question MUST include "Other — I'll type it" as the last option.
The predefined choices are starting points, not a cage. When the user picks
"Other — I'll type it" on any question, immediately follow up in plain text:
"What are you looking for?" and use their exact free-text answer in the profile.
Never assume or rephrase — store their words verbatim.

Always ask these 4 questions. **Build question 1 dynamically from the CV** — offer the
candidate's current-level role and the natural next step up, based on what the CV shows.

1. **Target role** (single select) — generate options from the candidate's CV:
   - [Their current role title — "stay at my level, better company"]
   - [The natural next step up from their current role]
   - [A lateral or adjacent role their background supports, if relevant]
   - Open to a range of roles at my level
   - Other — I'll type it

   Example for a "Senior Product Manager": options would be Senior Product Manager,
   Group Product Manager / Lead PM, Director of Product, Open to a range, Other.
   Example for a "Frontend Developer": options would be Frontend Developer,
   Senior Frontend Developer, Full-stack Developer, Open to a range, Other.

2. **Location preference** (single select):
   - My current city only
   - My metro area / commutable distance
   - Anywhere in my country
   - Open to remote
   - Other — I'll type it

3. **Company stage preference** (single select):
   - Pre-IPO startup (max equity upside)
   - Public company (stability + liquidity)
   - Growth stage Series B–D
   - Any — just show me the best fit
   - Other — I'll type it

4. **Preferred industry / domain** (single select) — adapt the examples to the
   candidate's field, but always keep these broad buckets:
   - Same industry as my current company
   - B2B SaaS / software
   - A specific industry I'll name
   - Any industry — fit matters more than sector
   - Other — I'll type it

**After the widget:** for every question where the user chose "Other — I'll type it",
ask that follow-up in plain text before proceeding to Step 1C.
Collect all answers (predefined + free text), then build the profile.

### Step 1C — Build and Confirm Candidate Profile

Combine CV data + user answers into a structured profile. Show the user a 5-line summary
and ask: "Does this look right?" before proceeding.

Store the profile — it will be used for ALL subsequent company rankings and interview prep.

Profile fields to store:
```
name, current_title, current_company, profession, seniority_level, years_exp,
skills[], tools[], domain_expertise[], managed_team_size, key_achievements[],
target_titles[], target_location, preferred_stage, preferred_industry,
education, notable_background
```

---

## Mode 2 — Company Ranking

Triggered when: user shares ANY company URL or company name after profile is set up.
This is the core daily-use mode. The user just drops a URL — the skill does everything.

From galorr/claude-code-starter, MIT.

More from galorr/claude-code-starter