/red-team-prd Adversarial Testing Methodology: How to Stress-Test PRDs and Roadmaps

The /red-team-prd command orchestrates the Strategy Red-Team skill to systematically identify load-bearing assumptions in product requirements documents, steel-man them into their strongest form, and generate ranked failure modes with actionable kill criteria and cheap tests.

The /red-team-prd command in the phuryn/pm-skills repository provides a structured adversarial testing framework for product managers to validate strategic plans before committing resources. Implemented as part of the pm-execution plugin, this command leverages the Strategy Red-Team skill to transform vague concerns into concrete, falsifiable risks. Unlike generic critique methods, it focuses exclusively on load-bearing assumptions—claims whose falsity would collapse the entire plan—and ranks them by a composite metric of impact, likelihood, and testing cost.

The Core Adversarial Testing Methodology

The methodology defined in /pm-execution/skills/strategy-red-team/SKILL.md and orchestrated via /pm-execution/commands/red-team-prd.md consists of five tightly-coupled stages:

Plan Ingestion and Context Parsing

The command accepts strategic input through three flexible interfaces. Users can provide a concise one-line hypothesis, paste or upload a full PRD or roadmap document, or reference "the current doc" when operating within an existing thread context. This ingestion stage normalizes any textual representation of a plan into a format suitable for analysis.

Assumption Extraction and Load-Bearing Filtering

Once ingested, the skill parses the plan to extract every underlying claim. It applies a critical filter: identifying only load-bearing assumptions—those specific claims whose falsity would cause the entire strategy to collapse. Cosmetic or secondary concerns are discarded to maintain focus on existential risks. This filtering ensures subsequent analysis targets only the structural vulnerabilities of the plan.

Steel-Man Construction and Rigorous Attack

For each load-bearing assumption identified, the methodology requires constructing the steel-man—the strongest possible justification for that claim. The adversarial attack then targets this fortified version rather than a straw-man weak interpretation. This approach, documented in the skill implementation, ensures the stress-test represents the most rigorous possible critique of the plan's actual logic.

Failure-Mode Formulation and Prioritization

Each successful attack is translated into a concrete, falsifiable failure mode formatted as "Fails if...". These failure modes are scored and ranked using the formula impact × likelihood × cheapness-to-test. The top 3-5 results surface the highest-impact, most-plausibly-wrong, yet cheapest-to-test risks first.

For each ranked failure mode, the output includes:

• Evidence to get this week: Specific data or conversations needed to confirm or refute the claim
• Kill criterion: The quantitative threshold at which the plan should be halted or pivoted
• Cheapest test: The smallest experiment capable of moving belief on the assumption

Output Formatting and Follow-Up Actions

The command formats results in a "screenshot-native" block optimized for immediate sharing. Following delivery, the system offers specific follow-up actions including converting kill-assumptions into formal experiments, running a pre-mortem analysis, or rewriting high-risk sections of the document.

How to Invoke the /red-team-prd Command

The command supports three input modalities as implemented in /pm-execution/commands/red-team-prd.md:

Quick hypothesis testing:

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

Full document analysis:

/red-team-prd [paste or upload a PRD, roadmap, or strategy]

Context-aware invocation using existing thread content:

/red-team-prd the current doc

Output Structure and Example

A typical response follows this structured format, designed for immediate action:

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

### Top Kill-Assumptions (ranked)

- **Claim:** Activation is the single constraint on growth
  - **Fails if:** Users churn before completing onboarding
  - **Evidence to get this week:** Cohort analysis of activation funnel drop-off
  - **Kill criterion:** >30% churn before step 3
  - **Cheapest test:** Run a 2-week A/B test of a shortened onboarding flow

### What's Well-Reasoned

[...]

Summary

• The /red-team-prd command leverages the Strategy Red-Team skill in the phuryn/pm-skills repository to perform rigorous adversarial testing of product strategies.
• It filters for load-bearing assumptions only, ignoring cosmetic concerns to focus on existential risks.
• The methodology employs steel-manning to ensure attacks target the strongest possible version of each claim.
• Failure modes are ranked by impact × likelihood × cheapness-to-test, prioritizing high-stakes, easily testable risks.
• Each output includes actionable kill criteria and cheapest tests to enable immediate validation work.

Frequently Asked Questions

What makes the /red-team-prd methodology different from standard risk analysis?

Traditional risk analysis often catalogs every possible concern without prioritization. The red-team-prd methodology specifically isolates load-bearing assumptions—those whose falsity collapses the entire plan—and ranks them by a composite score of impact, likelihood, and testing cost. This ensures teams address only the critical vulnerabilities that threaten strategic viability, rather than wasting time on peripheral issues.

How does the steel-man approach improve adversarial testing?

Rather than attacking weak or superficial interpretations of a plan (straw-manning), the methodology requires building the strongest possible justification for each load-bearing assumption before attempting to falsify it. According to the implementation in /pm-execution/skills/strategy-red-team/SKILL.md, this steel-man approach ensures that only genuinely robust strategies survive scrutiny, eliminating false confidence derived from easily demolished weak arguments.

What qualifies as a "load-bearing" assumption in this framework?

A load-bearing assumption is any claim within a PRD, roadmap, or strategy whose falsity would cause the entire plan to collapse or require fundamental restructuring. The parsing logic in the Strategy Red-Team skill specifically filters for these structural dependencies, discarding cosmetic or secondary concerns that might be wrong without invalidating the core strategy. This distinction allows teams to focus validation efforts exclusively on existential risks.

Can the /red-team-prd command analyze documents other than PRDs?

Yes. While named for product requirement documents, the command accepts any textual strategic plan including roadmaps, strategic memos, or one-line hypotheses. The ingestion logic in /pm-execution/commands/red-team-prd.md handles plain text, uploaded documents, or references to "the current doc" within an existing conversation thread, making it adaptable to various planning formats and contexts.

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