agentic-job-search-vault

内容来源：README.md（说明文档） · 原始地址 · 查看安装指南

原始内容

# Agentic Job Search Vault

An AI-powered workflow for high-impact job applications. Built for as an agentic Obsidian-style vault.

## What This Does

| Feature | Description |
|---------|-------------|
| **No-Slop Documents** | CVs and cover letters that sound human. Short sentences. Punchy rhythm. No corporate buzzwords. Ready in 2-3 prompts. |
| **Research-First Approach** | The agent researches the company before generating documents. Context makes tailoring precise. |
| **Browser Automation** | Built-in MCP for filling multi-page application forms (Workday, Greenhouse, Lever). You review. Then submit. |

## Requirements

- [opencode](https://opencode.ai/) (AI-powered terminal IDE)
- Chrome + [Playwriter Extension](https://chromewebstore.google.com/detail/playwriter-mcp/jfeammnjpkecdekppnclgkkffahnhfhe) (for form automation)

<img width="1024" height="559" alt="image" src="https://github.com/user-attachments/assets/d9976279-cbc8-4365-8582-465a943c3a35" />

## Quick Start

### 1. Clone and Open

```bash
git clone https://github.com/igor-warzocha/agentic-job-search-vault.git
cd agentic-job-search-vault
opencode
```

### 2. Add Your Materials

Drop everything you have into `01-Core-Materials/`. The more you give, the better the output.

| Material | Folder | Why It Helps |
|----------|--------|--------------|
| All your CVs | `01-Core-Materials/CVs/` | Different versions show range |
| Cover letters | `01-Core-Materials/Cover-Letters/` | Captures your voice |
| LinkedIn export | `01-Core-Materials/Portfolio/` | Rich context, endorsements |
| Project write-ups | `01-Core-Materials/Portfolio/` | Technical depth |
| Performance reviews | `01-Core-Materials/Portfolio/` | How others describe your impact |
| Old job descriptions | `01-Core-Materials/Portfolio/` | What you were hired to do |

**Formats:** Markdown, PDF, Word, plain text. Anything works.

### 3. Run Setup

```
/setup
```

The setup agent will:
1. Read everything in `01-Core-Materials/`
2. Build your `CANDIDATE-PROFILE.md` automatically
3. Ask clarifying questions
4. Configure agents with your preferences

You do NOT fill in `CANDIDATE-PROFILE.md` manually. The agent does it.

### 4. Install Playwriter (Optional)

For automating long application forms:

1. Install the [Playwriter Chrome Extension](https://chromewebstore.google.com/detail/playwriter-mcp/jfeammnjpkecdekppnclgkkffahnhfhe)
2. Pin it to your toolbar
3. Click the icon on any application page (green = connected)

The MCP config is already in `opencode.json`.

## Workflow

Once set up, the cycle is simple:

**Research** → Share a job posting. The agent researches the company first.

**Generate** → Tailored CV and cover letter created in `02-Applications/YYYY-MM/Company-Name/`.

**Apply** → Use Playwriter to fill forms. Review. Submit.

**Prep** → Interview materials generated in `04-Interview-Prep/`.

## Repository Structure

```
01-Core-Materials/          # YOUR INPUT: Past CVs, letters, portfolio
02-Applications/            # OUTPUT: Generated application packages
03-Job-Market-Research/     # Company research and market data
04-Interview-Prep/          # Interview preparation materials
CANDIDATE-PROFILE.md        # Auto-generated profile (source of truth)
```

## Style Rules

All generated documents follow strict quality standards:

- **No AI Slop:** No "leverage synergies" or "passionate about". Human language only.
- **British English:** UK spelling throughout. (ask the agent to change before you customise, if needed.)
- **Rich Conciseness:** Maximum 2 pages. Every word earns its place.

## License

MIT

Agentic Job Search Vault

An AI-powered workflow for high-impact job applications. Built for as an agentic Obsidian-style vault.

What This Does

Feature	Description
No-Slop Documents	CVs and cover letters that sound human. Short sentences. Punchy rhythm. No corporate buzzwords. Ready in 2-3 prompts.
Research-First Approach	The agent researches the company before generating documents. Context makes tailoring precise.
Browser Automation	Built-in MCP for filling multi-page application forms (Workday, Greenhouse, Lever). You review. Then submit.

Requirements

opencode (AI-powered terminal IDE)
Chrome + Playwriter Extension (for form automation)

Quick Start

1. Clone and Open

git clone https://github.com/igor-warzocha/agentic-job-search-vault.git
cd agentic-job-search-vault
opencode

2. Add Your Materials

Drop everything you have into 01-Core-Materials/. The more you give, the better the output.

Material	Folder	Why It Helps
All your CVs	`01-Core-Materials/CVs/`	Different versions show range
Cover letters	`01-Core-Materials/Cover-Letters/`	Captures your voice
LinkedIn export	`01-Core-Materials/Portfolio/`	Rich context, endorsements
Project write-ups	`01-Core-Materials/Portfolio/`	Technical depth
Performance reviews	`01-Core-Materials/Portfolio/`	How others describe your impact
Old job descriptions	`01-Core-Materials/Portfolio/`	What you were hired to do

Formats: Markdown, PDF, Word, plain text. Anything works.

3. Run Setup

/setup

The setup agent will:

Read everything in 01-Core-Materials/
Build your CANDIDATE-PROFILE.md automatically
Ask clarifying questions
Configure agents with your preferences

You do NOT fill in CANDIDATE-PROFILE.md manually. The agent does it.

4. Install Playwriter (Optional)

For automating long application forms:

Install the Playwriter Chrome Extension
Pin it to your toolbar
Click the icon on any application page (green = connected)

The MCP config is already in opencode.json.

Workflow

Once set up, the cycle is simple:

Research → Share a job posting. The agent researches the company first.

Generate → Tailored CV and cover letter created in 02-Applications/YYYY-MM/Company-Name/.

Apply → Use Playwriter to fill forms. Review. Submit.

Prep → Interview materials generated in 04-Interview-Prep/.

Repository Structure

01-Core-Materials/          # YOUR INPUT: Past CVs, letters, portfolio
02-Applications/            # OUTPUT: Generated application packages
03-Job-Market-Research/     # Company research and market data
04-Interview-Prep/          # Interview preparation materials
CANDIDATE-PROFILE.md        # Auto-generated profile (source of truth)

Style Rules

All generated documents follow strict quality standards:

No AI Slop: No "leverage synergies" or "passionate about". Human language only.
British English: UK spelling throughout. (ask the agent to change before you customise, if needed.)
Rich Conciseness: Maximum 2 pages. Every word earns its place.

License

MIT