What Programming Languages and Technologies Are Used in pm-product-discovery Skills?

The pm-product-discovery plugin relies entirely on Markdown with YAML front-matter and JSON configuration files—no traditional programming languages like Python, JavaScript, or Go are used.

The phuryn/pm-skills repository contains a collection of AI-assisted product management tools. The pm-product-discovery module is a unique, code-free plugin that defines behavior through structured text files interpreted by Claude-based AI assistants.

Technology Stack Overview

Unlike conventional software projects, pm-product-discovery contains zero compiled source code. The entire "technology stack" consists of declarative text formats that describe skills, commands, and plugin metadata.

Markdown with YAML Front-Matter

Each skill resides in its own SKILL.md file. For example, pm-product-discovery/skills/prioritize-features/SKILL.md contains a YAML front-matter header wrapped in triple dashes (---), followed by Markdown content that defines the skill's behavior for Claude.

---
name: prioritize-features
description: Evaluate features against impact, effort, risk, and strategic alignment
---

# Prioritize Features Skill

Analyze the feature backlog and return the top 5 recommendations...

This format provides Claude with the metadata required to load the skill and present it to users, while the Markdown body describes the execution logic in natural language instructions.

JSON Plugin Manifest

The file pm-product-discovery/.claude-plugin/plugin.json serves as the plugin's configuration manifest. It supplies plugin-level metadata including name, version, author, and keywords required by Claude's plugin marketplace integration.

{
  "name": "pm-product-discovery",
  "version": "1.0.0",
  "author": "phuryn",
  "keywords": ["product-management", "discovery"]
}

Plain-Text Distribution Format

The repository distributes as a plain-text archive with no binary assets. This allows the plugin to be consumed by Claude Code, Claude Cowork, OpenCode, Gemini CLI, Cursor, Kiro, and other compatible environments simply by copying the Markdown files into the appropriate skills folder.

Key Files in the Repository

Understanding the file structure reveals how the technology choices support the plugin's architecture:

• pm-product-discovery/skills/prioritize-features/SKILL.md — Feature-prioritization logic using impact × effort matrices
• pm-product-discovery/skills/summarize-interview/SKILL.md — Interview-summarization templates with JTBD-focused outputs
• pm-product-discovery/skills/brainstorm-ideas-existing/SKILL.md — Ideation workflows for existing products
• pm-product-discovery/skills/identify-assumptions-existing/SKILL.md — Assumption-mapping for risk identification
• pm-product-discovery/commands/discover.md — Orchestrates full discovery workflows by chaining multiple skills
• pm-product-discovery/commands/brainstorm.md — Exposes ideation capabilities as slash-commands
• pm-product-discovery/.claude-plugin/plugin.json — Plugin metadata for Claude's marketplace integration

How Skills Are Executed

The pm-product-discovery skills are language-agnostic descriptions interpreted by the Claude AI model. Users invoke functionality through slash-commands defined in Markdown files within the commands/ directory.

For example, the /discover command defined in pm-product-discovery/commands/discover.md chains several skills:

/discover AI-powered meeting summarizer for remote teams

This command sequentially executes:

1. brainstorm-ideas-* — Ideation of product concepts
2. identify-assumptions-* — Surfacing risky assumptions
3. prioritize-assumptions — Ranking by impact × risk
4. brainstorm-experiments-* — Designing validation experiments

Similarly, /prioritize-features invokes the prioritize-features skill to return the top 5 recommendations based on strategic alignment, while /interview summarize activates the summarize-interview skill to structure raw transcripts into JTBD-focused summaries.

Summary

• Zero compiled code: The pm-product-discovery plugin uses only Markdown and JSON, making it runtime-dependency free
• YAML front-matter: Supplies metadata in SKILL.md files for Claude's skill loader
• JSON manifest: plugin.json registers the plugin with Claude's marketplace system
• Language agnostic: Skills work with any AI assistant that understands the Claude skill schema
• Plain-text portability: Files copy directly into Claude Code, Cowork, Cursor, and other compatible environments without build steps

Frequently Asked Questions

Does pm-product-discovery use Python or JavaScript?

No. According to the phuryn/pm-skills source code, there is no Python, JavaScript, Go, or any other compiled programming language in the discovery plugin. The skills are defined entirely in Markdown files with YAML headers that Claude interprets directly as natural language instructions.

What is the role of the plugin.json file?

The pm-product-discovery/.claude-plugin/plugin.json file contains the plugin manifest required by Claude's plugin system. It declares the plugin name, version, author, and keywords, enabling the plugin to be recognized and loaded by Claude Code and Claude Cowork.

How do I install pm-product-discovery skills in Claude Code?

Because the technology stack consists solely of plain-text Markdown and JSON files, installation requires no package managers or compilers. Simply copy the skill files into your Claude environment's skills folder, and the YAML front-matter in files like prioritize-features/SKILL.md will automatically register the capabilities with the AI assistant.

Can I use these skills with other AI assistants besides Claude?

Yes. The Markdown/YAML format is designed to be language-agnostic. While optimized for Claude-based systems (Claude Code, Claude Cowork), the structured text can be consumed by any AI assistant that understands the Claude skill schema, including OpenCode, Gemini CLI, Cursor, and Kiro.

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