AI Call

A skill that turns any AI coding agent into a structured interviewer. The agent asks questions one at a time, follows up based on your answers, and produces a clean markdown transcript.

Works with any agent that supports the Agent Skills standard: pi, Claude Code, OpenAI Codex, and others.

What It Does

1. You provide a prompt — a topic, questions, or a job description
2. The AI asks questions one at a time, waiting for your response
3. It follows up if your answer is incomplete or opens an interesting thread
4. When done, it generates a clean markdown transcript with all Q&A pairs

No API keys needed — uses whatever model your agent is already running.

Install

Pi

# From npm
pi install npm:ai-call

# From git
pi install https://github.com/nicola/ai-call

# Pin a version
pi install npm:ai-call@1.0.0
pi install https://github.com/nicola/ai-call@v1.0.0

Claude Code

# Global (available in all projects)
git clone https://github.com/nicola/ai-call ~/.claude/skills/ai-call

# Or project-level
git clone https://github.com/nicola/ai-call .claude/skills/ai-call

OpenAI Codex

# Global
git clone https://github.com/nicola/ai-call ~/.agents/skills/ai-call

# Or project-level
git clone https://github.com/nicola/ai-call .agents/skills/ai-call

Manual

Copy SKILL.md into your agent's skills directory. That's the only file that matters.

Usage

Once installed, just ask your agent:

• "Start an AI call about my experience with distributed systems"
• "Interview me about my leadership style"
• "Ask me questions about this job description: [paste JD]"
• "Run an AI call — ask about my top 3 projects and what I learned"

The agent will ask questions one at a time, follow up naturally, and save a transcript as YYYY-MM-DD - [Topic].md.

Examples

Topic-based:

"Start an AI call about my experience building mobile apps"

Question list:

"Ask: 1) What's your biggest technical challenge? 2) How do you handle disagreements? 3) What are you learning right now?"

Job description:

"Interview me for this role: [paste job description]"

Deep dive:

"Ask about my distributed systems experience. Deep dive on consistency models and failure handling. Cover 3-4 real projects."

How It Works

This is a pure-prompt skill — no scripts, no dependencies, no API keys. The SKILL.md file contains instructions that tell the agent how to behave as an interviewer. Your agent loads these instructions when it detects a matching request.

The transcript is saved locally as a markdown file. Your data never leaves your machine (beyond whatever your agent's model provider already sees).