AI Job Search Framework Entry-Point Slash Commands: The Complete Reference

The AI Job Search Framework exposes twelve entry-point slash commands—defined as Markdown files in .claude/commands/—that serve as the canonical public API for onboarding candidates, ranking opportunities, managing applications, and syncing external data.

The MadsLorentzen/ai-job-search repository implements a conversational job-search automation system driven by these slash commands. Each command acts as a primary entry point for specific workflows, parsed by the framework's dispatcher to orchestrate underlying skills located in .claude/skills/.

Core Workflow Commands

The framework organizes job searching into a logical lifecycle, from initial onboarding through application submission.

/setup – Candidate Onboarding

The /setup command initiates the candidate profile creation process. Located in .claude/commands/setup.md, this entry point collects personal data, parses uploaded documents, and generates the core skill files required for personalized job matching.

When executed, the command populates the candidate profile stored in .claude/skills/job-application-assistant/ and establishes the initial state files.

/reset – Profile Management

Use /reset to clear candidate-profile data and optionally remove the documents/ folder. According to the source code in .claude/commands/reset.md, this command supports targeted resets (e.g., /reset profile) that preserve uploaded documents while clearing derived data, allowing users to start fresh without re-uploading source files.

/rank – Job Evaluation

The /rank command consumes scraped job postings and produces a ranked shortlist based on the personalized evaluation model defined in the candidate's skill files. Implemented in .claude/commands/rank.md, this command compares job requirements against the competency matrix stored in the framework's state.

/apply – Application Execution

/apply executes the full application workflow. As specified in .claude/commands/apply.md, this command drafts cover letters, tailors the CV using registered templates, and prepares the final submission packet. This represents the primary action command in the application phase.

/outcome – Result Tracking

Record application results using /outcome. The handler in .claude/commands/outcome.md accepts statuses such as hired, rejected, or interview-only, feeding this data back into the evaluation loop to refine future ranking algorithms.

Data Integration and Reporting Commands

These entry points manage external synchronization and reporting functions.

/expand – Competency Extraction

The /expand command extracts additional competency data from uploaded documents or online profiles. Defined in .claude/commands/expand.md, it augments the candidate profile by parsing new information sources without requiring a full reset.

/notion-sync – External Database Integration

/notion-sync pushes ranked jobs and application updates to a Notion database for external tracking. The implementation in .claude/commands/notion-sync.md handles API authentication and schema mapping between the framework's internal state and Notion pages.

/gmail-sync – Email Status Integration

Synchronize application status changes from a Gmail mailbox using /gmail-sync. According to .claude/commands/gmail-sync.md, this command scans designated mailboxes for recruiter responses and updates the internal tracker automatically.

/html-report – Dashboard Generation

Generate visual summaries with /html-report. The command defined in .claude/commands/html-report.md builds an HTML dashboard summarizing the application tracker, search results, and pipeline status for quick overview.

/interview – Preparation Material

The /interview command produces interview preparation content. As implemented in .claude/commands/interview.md, it generates STAR-method candidate examples and question banks based on the specific job requirements of tracked applications.

Extensibility Commands

These entry points allow customization of the framework's capabilities.

/add-template – Template Registration

Register custom CV or cover-letter templates using /add-template. The handler in .claude/commands/add-template.md validates template files (e.g., /add-template tech-lead.tex) and registers them for reuse in future /apply workflows.

/add-portal – Portal Skill Generation

/add-portal creates new search skills for job markets not supported out-of-the-box. According to .claude/commands/add-portal.md, accepting a URL argument (e.g., /add-portal https://www.seek.com.au), this command generates portal-specific CLI tools in .agents/skills/.

Command Invocation Syntax

All slash commands are invoked from the repository root directory. The framework parses the $ARGUMENTS variable to handle parameters.

Initialize a fresh profile:

/setup

Reset only profile data while preserving documents:

/reset profile

Rank recently scraped jobs:

/rank

Record an interview outcome:

/outcome interview

Sync Gmail and generate reports:

/gmail-sync
/html-report

Add new resources:

/add-template tech-lead.tex
/add-portal https://www.seek.com.au

File Architecture

The command system relies on a specific directory structure that separates entry points from implementation:

• Command definitions: .claude/commands/*.md — Contains all 12 entry-point files, each beginning with # /<command> and documenting workflow steps.

• Core skills: .claude/skills/job-application-assistant/*.md — Stores candidate profiles, behavioral data, evaluation models, and templates.

• Search configuration: .claude/skills/job-scraper/search-queries.md — Holds personalized query snippets consumed by scraping workflows.

• Portal implementations: .agents/skills/*/cli/ — Houses portal-specific CLIs (e.g., linkedin-search, jobindex-search) implementing search and detail sub-commands.

• Documentation: README.md, SETUP.md, AGENTS.md — Provide project overview and agent catalogues.

These components form the complete public API: the Markdown-defined commands provide the interface, while the skills and agents provide the implementation.

Summary

• The AI Job Search Framework provides twelve entry-point slash commands defined as Markdown files in .claude/commands/.
• Commands follow a lifecycle pattern: setup (/setup, /reset), execution (/rank, /apply), tracking (/outcome), and integration (/notion-sync, /gmail-sync).
• Extensibility is built-in via /add-template and /add-portal, allowing customization without code modification.
• All commands parse $ARGUMENTS for parameters and update shared state files including CLAUDE.md and the documents/ directory.

Frequently Asked Questions

How do I invoke entry-point slash commands in the AI Job Search Framework?

Enter the command directly in the conversational interface or terminal, prefixed with a forward slash. The dispatcher recognizes the pattern /<command> and routes execution to the corresponding .claude/commands/<command>.md file.

What is the difference between /reset and /setup?

The /setup command performs initial candidate onboarding and creates profile files from scratch. The /reset command selectively clears existing data—optionally preserving uploaded documents in the documents/ folder—allowing you to reconfigure the profile without losing source materials.

Where are slash command definitions stored?

All command definitions reside in .claude/commands/ as Markdown files. Each file begins with a title of the form # /<name> – <description> and contains the step-by-step procedure the dispatcher executes.

Can I create custom slash commands for new job portals?

Yes. Use the /add-portal command followed by the portal URL. This generates a new portal-specific search skill in .agents/skills/, enabling the framework to scrape and rank jobs from that source without manual code changes.

{
  "mcpServers": {
    "instagit": {
      "command": "npx",
      "args": ["-y", "instagit@latest"]
    }
  }
}