What Is the Strategy-Red-Team Skill and How Does It Perform Adversarial Stress-Testing?

The strategy-red-team skill is a built-in PM-Skills tool that adversarially stress-tests product plans by attacking their load-bearing assumptions to surface concrete failure modes and rank them by impact, likelihood, and cheapness-to-test.

The strategy-red-team skill is a first-class capability in the phuryn/pm-skills repository designed to help product managers rigorously validate strategic documents. Unlike generic risk checklists, this tool programmatically adopts the mindset of a hostile adversary to identify which assumptions must hold true for your plan to succeed, then systematically attempts to break them. By integrating this skill into your workflow via the /red-team-prd command, you convert speculative planning into an evidence-based validation process.

How the Strategy-Red-Team Skill Works

The adversarial stress-testing workflow is defined in pm-execution/skills/strategy-red-team/SKILL.md and follows a disciplined five-step process that mirrors professional red-team exercises.

Extract Load-Bearing Assumptions

The skill first parses your document to distinguish between load-bearing assumptions (claims that must be true for the plan to succeed) and cosmetic assertions. It ignores fluff and focuses exclusively on the structural supports of your strategy.

Steel-Man and Attack

Rather than constructing straw-man arguments, the skill steel-mans each claim—stating the strongest possible justification for it—then deliberately attacks that robust version. This ensures the stress-test targets the most defensible version of your logic, not a weak mischaracterization.

Write Falsifiable Failure Modes

Each identified risk is expressed as a concrete, testable statement following the format: "Fails if…". This makes every assumption falsifiable and eliminates vague "might happen" language from the output.

Rank by Risk Score

Assumptions are ranked using the product of impact × likelihood × cheapness-to-test. The highest-ranked items represent the most dangerous risks that are also easiest to validate, ensuring you prioritize experiments that provide the highest signal-to-cost ratio.

Deliver Actionable Kill Criteria

For each top kill-assumption, the skill returns four specific fields:

• Fails if – The precise condition that invalidates the assumption
• Evidence to get this week – The data or conversation needed to confirm or kill it
• Kill criterion – The threshold at which to abandon or revise the plan
• Cheapest test – The smallest experiment that validates the hypothesis

Where to Find the Implementation

The strategy-red-team capability is implemented across three key locations in the repository:

• Skill definition: pm-execution/skills/strategy-red-team/SKILL.md contains the detailed instructions, constraints, and output formatting rules that govern the adversarial analysis.
• Command wrapper: pm-execution/commands/red-team-prd.md defines the user-facing /red-team-prd command that invokes the skill.
• Repository index: The root README.md lists the skill as a first-class capability: "strategy-red-team — Adversarial stress-test of a plan".

How to Use the Red-Team-PRD Command

The /red-team-prd command provides a seamless interface to the strategy-red-team skill. You can invoke it with a full document, a one-line hypothesis, or by referencing your current context.

# Example 1 – Red-team a full PRD

/red-team-prd
Paste your product requirements document (or upload a file)

# Example 2 – Red-team a strategic hypothesis

/red-team-prd Prioritize AI onboarding — activation is our bottleneck

# Example 3 – Use current document context

/red-team-prd the current doc

When executed, the system loads the supplied text, executes the strategy-red-team skill, and returns a formatted report. For example, analyzing the hypothesis "Prioritize AI onboarding — activation is our bottleneck" produces:

## Red-Team: Prioritize AI onboarding — activation is our bottleneck

### Top Kill‑Assumptions (ranked)

- **Claim:** Activation is the primary constraint on growth
  - **Fails if:** Activation metrics stay flat despite increased spend
  - **Evidence to get this week:** Run a 48‑hour AB test on a new onboarding flow
  - **Kill criterion:** < 5 % lift in activation over baseline
  - **Cheapest test:** Deploy a feature flag and measure activation lift

### What's Well‑Reasoned

... (explicitly states what the plan already supports)

### What I Couldn't Assess

... (gaps where the doc lacked data)

The skill enforces strict discipline: it never fabricates risks, explicitly marks what is well-reasoned versus uncertain, and avoids generic risk lists. The output is optimized for screenshot-sharing and includes optional prompts for converting kill-assumptions into experiments.

Summary

• The strategy-red-team skill is located in pm-execution/skills/strategy-red-team/SKILL.md and invoked via the /red-team-prd command defined in pm-execution/commands/red-team-prd.md.
• It performs adversarial stress-testing by steel-manning and then attacking load-bearing assumptions rather than cosmetic claims.
• Risks are ranked by impact × likelihood × cheapness-to-test, prioritizing high-danger, low-cost validations.
• Output includes falsifiable "Fails if" statements, kill criteria, and cheapest tests to guide immediate validation.
• The skill explicitly distinguishes between well-reasoned claims and gaps it cannot assess, ensuring no hallucinated risks enter the analysis.

Frequently Asked Questions

What makes the strategy-red-team skill different from a standard risk analysis?

The strategy-red-team skill in pm-execution/skills/strategy-red-team/SKILL.md specifically targets load-bearing assumptions rather than surface-level risks, and it requires every risk to be expressed as a falsifiable "Fails if" statement. Unlike generic risk matrices, it ranks threats by the product of impact, likelihood, and cheapness-to-test, ensuring you tackle the most dangerous assumptions that are cheapest to validate first.

How does the skill determine which assumptions to test first?

According to the skill specification, the ranking algorithm multiplies impact × likelihood × cheapness-to-test. This formula prioritizes assumptions that would catastrophically break your plan if wrong, are likely to be false, and can be tested with minimal resources. The highest-ranked items appear at the top of your kill-assumption report with specific "Evidence to get this week" recommendations.

Can I use the strategy-red-team skill on documents other than PRDs?

Yes. While the command is named /red-team-prd, the strategy-red-team skill accepts any strategic document, roadmap, or even one-line hypotheses. The README.md describes it broadly as an "Adversarial stress-test of a plan," and the command wrapper in pm-execution/commands/red-team-prd.md supports text inputs ranging from full documents to brief strategic statements.

What file contains the detailed instructions for the red-team analysis?

The complete behavioral specification, including the five-step workflow and output formatting rules, resides in pm-execution/skills/strategy-red-team/SKILL.md. This file instructs the AI to steel-man claims, write concrete failure modes, and structure responses with specific fields like "Kill criterion" and "Cheapest test."

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