Strategy red-team skill for Claude: stress-test a plan before you ship it

Makes Claude red-team a PRD, roadmap or strategy: it finds the assumptions that would sink it and gives the cheapest test for each.

Installs: 9.4K; on skills.sh
Stars: 27.6K; on GitHub
Updated: Oct 2026; last repo commit

Claude lists what your plan claims is true, keeps only the load-bearing claims, states the strongest version of each, then attacks that version. Every failure mode is written as 'Fails if ___' and ranked by impact, likelihood of being wrong and cheapness to test.

For each surviving assumption you get the evidence to collect this week, a kill criterion and the cheapest test. The answer also says what holds up and what Claude could not assess. SKILL.md is one file with no scripts.

Good for

Product managers preparing a PRD or roadmap for executive review
Founders and leads who want a hard look at a strategy bet
Teams that want a test and a stop rule for each risky assumption

Install strategy-red-team

Claude Code

Copy this prompt.

Install the strategy-red-team skill in Claude Code. Run this command:

claude plugin marketplace add phuryn/pm-skills
claude plugin install pm-execution@pm-skills

Then check that it loaded and tell me how to use it.

2
Paste it into Claude Code. It shows the change and asks before running it.

Pick one route. The plugin route installs the pm-execution plugin: 16 skills and 11 commands (counted from the repo tree, with no hooks, agents or MCP config files in the plugin folder). The README names a separate Claude Cowork route that installs all 9 plugins. The skills CLI and folder routes install just this skill. The /red-team-prd command comes only with the plugin.

From the project README. Source.

What's in the skill

When it loads

Red-team a PRD, roadmap, or strategy by attacking its load-bearing assumptions before reality does. Steelmans then attacks each claim, ranks failure modes by impact × likelihood × cheapness-to-test, and returns the cheapest test and kill criteria for each. Use when stress-testing a plan, pressure-testing a strategy, challenging assumptions, or preparing a doc for executive review.

Sections in SKILL.md

Purpose
Context
Instructions
Notes
Further Reading

Key facts

Author: Paweł Huryn (named in the plugin.json of pm-execution)
License: MIT
Installs on skills.sh: 9,382 (Nov 2, 2026)
GitHub stars: 27,593 for the whole phuryn/pm-skills repo (Nov 2, 2026)
Last repo commit: Oct 29, 2026
Audits on skills.sh: Gen Agent Trust Hub: pass, Socket: pass, Snyk: pass
What it includes: 1 file: SKILL.md (bundles no scripts)
Part of: The pm-execution plugin: 16 skills and 11 commands (counted from the repo tree)
Command: /red-team-prd runs this skill, per the command file

What it includes

Lists what the plan says is true
Separates load-bearing claims (if false, the plan dies) from cosmetic ones. Only load-bearing claims get attacked.
Try: “Red-team this PRD: we will prioritize AI onboarding because activation is our bottleneck.”
Read · 1. Extract every claim
Strongest case for each claim, then the attack
No strawmen: the attack targets the strongest version of the claim.
Read · 2. Steelman, then attack
Concrete, falsifiable failure modes
Each failure mode is one condition that can be checked, not a general risk like execution risk.
Read · 3. Write "Fails if ___"
Impact times likelihood wrong times cheapness to test
The top of the list is what to test this week.
Read · 4. Rank
Says plainly what is well reasoned
SKILL.md says never to invent a weakness the plan does not have.
Read · 5. Self-refute
Fails if, evidence this week, kill criterion, cheapest test
Four fields for every surviving assumption, then an output with 3 to 5 top kill-assumptions, what is well reasoned and what could not be assessed.
Read · 6. Act on each kill-assumption
Second opinion from another model, only if you ask
If you ask and another model such as Codex, Gemini or a second Claude is reachable, Claude runs the plan through it and flags disagreements. Single-model is the default.
Read · 7. Optional cross-model mode

From the official docs

Permissions

Low risk

What it can do

Read your data
Create and change things
Delete things
Read-only mode

Watch out for

Cross-model mode sends your plan to another model
SKILL.md step 7 says that if you ask for a second opinion and another model is reachable, Claude runs the same plan through it. It names Codex, Gemini and a second Claude and does not say how they are reached. Single-model is the default; ask for it only if you are happy to share the plan with that provider. github.com
The plugin adds a command
The pm-execution plugin adds the /red-team-prd command, which applies this skill to the plan you pass it. The skills CLI route installs only the skill. github.com

Security advisories

No published security advisories as of Nov 2, 2026.

Technical details

Turning on read-only	SKILL.md describes analysis and a written answer; it names no file edits.
Sign-in and keys	SKILL.md names no key or account.
Scopes	Reads the plan, PRD or document you give it In the optional cross-model mode, can send the same plan to another model such as Codex, Gemini or a second Claude

Questions

What does strategy-red-team do?

It makes Claude attack the load-bearing assumptions of a PRD, roadmap or strategy. For each one it gives a failure condition, the evidence to get this week, a kill criterion and the cheapest test.

How do I install it?

Use the install box at the top of this page. The plugin tab installs the pm-execution plugin, which also adds the /red-team-prd command and 15 other skills.

How is it different from a pre-mortem?

SKILL.md says a pre-mortem imagines the plan already failed and tells the story of why. The red-team attacks the live assumptions now and hands you the cheapest test. The same plugin has a separate pre-mortem skill.

What do I give it?

Paste or upload the plan: a PRD, roadmap, strategy memo or a one-line bet. The /red-team-prd command also accepts "the current doc".

Will it just find lots of generic risks?

SKILL.md says no generic risk lists: every item must be specific to your plan, five real kill-assumptions beat twenty generic ones, and it must say what is well reasoned.

Is it safe?

The skill is one text file with no scripts. It asks Claude to analyze the plan you give it, so the document you paste is shared with your AI provider like any other chat input.

Information

updated Nov 2, 2026

Repository: phuryn/pm-skills
Folder: pm-execution/skills/strategy-red-team

phuryn/pm-skills skills.sh

Sources

Docs, issue trackers and discussions this page is built from.

Show all 4

strategy-red-team SKILL.md · github.com
red-team-prd command · github.com
pm-skills README · github.com
pm-skills marketplace.json · github.com