How the /discover Command Chains Multiple Skills Together for Full Product Discovery

The /discover command orchestrates a seven-step workflow that chains together specialized skills—from ideation to experiment design—to deliver a complete product discovery plan.

The /discover command in the phuryn/pm-skills repository provides a structured approach to chain multiple skills together for end-to-end product discovery. This high-level workflow guides product teams from initial ideation through assumption validation to concrete experiment planning using modular, reusable AI prompts defined in Markdown.

The Seven-Step Discovery Pipeline

The command defined in pm-product-discovery/commands/discover.md implements a sequential pipeline where each stage feeds into the next. The workflow combines five specialized skills with two manual checkpoints to create a cohesive discovery experience.

Step 1: Context Gathering

The command initiates by asking the user to clarify the product context, existing knowledge, and specific decision goals. This step requires no external skill invocation—it operates directly within the command logic to establish the foundation for subsequent analysis.

Step 2: Idea Generation

The command invokes either brainstorm-ideas-existing or brainstorm-ideas-new depending on product maturity. These skills, defined in pm-product-discovery/skills/brainstorm-ideas-existing/SKILL.md and its counterpart, execute multi-perspective ideation across Product Management, Design, and Engineering viewpoints. Each skill returns ten distinct ideas for user selection.

Step 3: Assumption Mapping

Based on selected ideas, the command chains to identify-assumptions-existing or identify-assumptions-new. These skills surface critical assumptions across Value, Usability, Feasibility, and Viability dimensions. For new products, the analysis extends to include Go-to-Market (GTM) assumptions.

Step 4: Assumption Prioritization

The prioritize-assumptions skill plots surfaced assumptions on an Impact × Risk matrix. This step identifies "leap-of-faith" assumptions—high-risk, high-impact items that require immediate validation—and ranks them for experiment design.

Step 5: Experiment Design

For the highest-priority assumptions, the command invokes brainstorm-experiments-existing or brainstorm-experiments-new. These skills generate one to two concrete validation experiments per assumption, including A/B tests, pretotypes, landing pages, or concierge tests.

Step 6: Discovery Plan Assembly

The command compiles all outputs—selected ideas, mapped assumptions, and experiment specifications—into a comprehensive markdown Discovery Plan. The template for this document is defined in lines 81-114 of discover.md.

Step 7: Next-Step Suggestions

Finally, the command offers contextual follow-up actions such as PRD creation, interview-script generation, or metrics setup, allowing teams to transition directly from discovery into execution.

How the Pipeline Architecture Works

The /discover command functions as a pipeline where output from each skill becomes the input for the subsequent stage. This architecture creates seamless data flow while maintaining user control through strategic checkpoints.

Checkpoints appear after skill completion (e.g., "Here are 10 ideas…") allowing users to confirm outputs or adjust the flow before proceeding. This ensures the process remains interactive yet automated.

Each skill operates as a stand-alone, reusable Markdown-defined AI prompt. Because skills are independent modules stored in pm-product-discovery/skills/, the /discover command can be extended or re-ordered without modifying the underlying engine—simply by swapping skill references. This modular design makes the discovery workflow both composable and maintainable across the nine PM plugins provided by the repository.

Key Implementation Files

The following source files define the /discover command chain:

File	Role
pm-product-discovery/commands/discover.md	Orchestrates the end-to-end seven-step workflow
pm-product-discovery/skills/brainstorm-ideas-existing/SKILL.md	Generates ten product ideas for existing products
pm-product-discovery/skills/brainstorm-ideas-new/SKILL.md	Generates ten product ideas for new products
pm-product-discovery/skills/identify-assumptions-existing/SKILL.md	Extracts risk assumptions for existing products
pm-product-discovery/skills/identify-assumptions-new/SKILL.md	Extracts risk assumptions for new products
pm-product-discovery/skills/prioritize-assumptions/SKILL.md	Ranks assumptions on impact vs. uncertainty
pm-product-discovery/skills/brainstorm-experiments-existing/SKILL.md	Generates validation experiments for existing products
pm-product-discovery/skills/brainstorm-experiments-new/SKILL.md	Generates validation experiments for new products
README.md	Documents the command structure and skill chaining
.claude-plugin/marketplace.json	Lists the discovery plugin capabilities

Example Workflow Execution

Users invoke the command with a simple natural language prompt:

# Example invocation

/discover Smart notification system for our project management tool

The typical interaction flow follows this pattern:

1. Context – The command asks: "What are you exploring? …"
2. Brainstorm – The skill returns ten ideas; the command prompts: "Pick 3-5 to carry forward."
3. Assumptions – The skill lists assumptions for each selected idea.
4. Prioritize – The skill produces an Impact × Risk matrix and highlights the top-risk assumptions.
5. Experiments – The skill suggests concrete tests (e.g., A/B test a notification banner).
6. Discovery Plan – The command assembles a markdown document using the template in discover.md.
7. Next steps – The command offers to generate a PRD, interview script, etc.

Summary

• The /discover command chains five specialized skills across seven sequential steps to deliver complete product discovery workflows.
• Skills are modular Markdown files stored in pm-product-discovery/skills/, making the pipeline extensible without engine changes.
• The workflow alternates between skill invocations (steps 2-5) and manual checkpoints (steps 1, 6, 7) to balance automation with user control.
• Impact × Risk matrices and leap-of-faith assumption identification drive prioritization before experiment design.
• The final output is a structured Discovery Plan compiled from all preceding steps, defined in lines 81-114 of pm-product-discovery/commands/discover.md.

Frequently Asked Questions

What is the difference between the "existing" and "new" skill variants in the /discover command?

The "existing" variants (brainstorm-ideas-existing, identify-assumptions-existing, brainstorm-experiments-existing) focus on iterative improvements to current products and analyze Value, Usability, Feasibility, and Viability. The "new" variants add Go-to-Market (GTM) assumption analysis and are optimized for zero-to-one product development.

How does the /discover command maintain user control between automated steps?

The command implements checkpoints after each skill completion, presenting outputs (such as the ten generated ideas or the Impact × Risk matrix) and explicitly prompting for user confirmation or adjustment before proceeding to the next skill in the chain.

Where is the Discovery Plan template defined in the source code?

The Discovery Plan template is embedded directly in the command definition at lines 81-114 of pm-product-discovery/commands/discover.md. This template specifies how to compile ideas, assumptions, and experiment specifications into the final markdown document.

Can the /discover workflow be customized to skip specific steps or rearrange the skill order?

Yes. Because each skill is a stand-alone, reusable Markdown-defined AI prompt, you can modify the skill chain in pm-product-discovery/commands/discover.md without touching the underlying execution engine. Simply reference different skill files or reorder the invocation sequence to customize the discovery workflow.

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