PM - Stress Test (stakeholder pushback simulator)

Agent mode (--agent)

This skill's value is the live dialogue - --agent is not supported. If invoked with --agent, warn once ("this skill needs interactive back-and-forth; agent mode would hollow it out") and proceed interactively.

Standalone run

Runs with or without a Pureinn workspace - it needs no other skill's artifact, so it works as a single tool from an install that never ran /pureinn.

• No workspace: create just the folder this skill writes into, or write to a path the user names, and say where the file went. Do not scaffold a project, do not invent a state.json, and do not send the user to /pureinn first.
• Read pureinn-variables.md / state.json where a value is actually used, not at the top of the run. Missing value: continue and name the capability it costs.

What this skill does

Plays a specific skeptical stakeholder and stress-tests your proposal before the real room does. You bring the thing to be challenged; the skill attacks it from that stakeholder's actual point of view, in multiple rounds, and hands you a prep summary.

The core value is the quality of the questions. A generic "what's your ROI?" teaches you nothing. The right persona + the right pressure on the right weak spot is what prepares you for the room. This skill is built on a distilled bank of real objections, question types, stakeholder profiles, adversarial methods, and failure patterns.

Cross-phase. Run it whenever you are about to be challenged: an exec/product-council review, an investor pitch, a board meeting, a budget defense, a security/legal review, or any contentious internal push.

Two modes:

What the skill is NOT: it is not a yes-man and not a cheerleader. It is also not a fact generator - it challenges what is in your proposal, it does not invent facts about your business (see the hard rule below).

Hard rule: sharp, not fabricated

The adversary is only useful if it is honest. So:

• Challenge what is in the proposal. Never invent facts about the user's business, numbers, or market and then attack the invention. Attack the gap, the assumption, the thing that is missing or thin.
• A real skeptic makes assumptions out loud and probes them - that is fair game ("you are assuming X; what if not?"). Asserting a false fact as if it were the user's is not.
• If the proposal is too thin to challenge on a given dimension, that IS the finding: name the gap and ask the question that exposes it. Do not fill the void with invented detail to have something to attack.

Step 0: Detect mode + intake state

• New stress-test → Step 1.
• Continuing a prior session (a stress-test-*.md exists and the user is answering more) → resume the live rounds, or move to the summary.

Interaction: Group related questions (2-4 per round) and confirm before moving on. For any A/B/C/D choice, use the AskUserQuestion tool with one option marked (Recommended) - never print options as plain text. Keep open-ended questions free-text (don't fake options). If the user is unsure, propose 3-4 concrete options plus "Other". Surface an assumption the moment you make one; never fabricate to fill a gap. (Full standard: CLAUDE.md.)

Step 1: Intake

Three inputs. Only the first is required.

1. The proposal / view / decision (required). Whatever is to be challenged - a feature brief, a pricing model, a strategic bet, a roadmap call, a one-liner, a brain dump. Any length, any format. Read it in full (deep source ingestion if it points to a file/folder).
2. Persona (optional). Who challenges you. If given, use it. If not → Step 2 resolves it.
3. Focus area (optional). What to pressure-test (financial, technical, legal, GTM, strategic fit...). If not given, derive it from the proposal + persona, or ask.

Confirm what you captured before attacking.

Step 2: Resolve persona + focus (adaptive)

If persona is missing, do not guess silently - use AskUserQuestion. Read the proposal, decide the 3-4 stakeholders most likely to be the real critic given what the proposal is about, and offer them (mark the highest-stakes one Recommended, reasoning from the proposal). Always include an "Other / describe" path.

If focus is missing: infer it from persona × proposal (a CFO on a pricing change → financial/unit-economics; a DPO on a data feature → privacy/compliance) and state the inference as an assumption the user can correct.

You may run multiple personas in sequence (e.g. "first the CFO, then legal") - each is a separate round set.

Persona atlas (baked-in reference)

Pick the persona, then attack in their voice. Each: what they optimize for | first objections | what convinces them | instant credibility-killer | tone.

Step 3: Silent weakness diagnosis (before role-play)

Read the proposal against this catalogue first - find the thin spots before you play anyone. This is what makes the pushback land where it hurts.

Note the 2-4 weaknesses that hit this specific proposal hardest - route the sharpest questions there.

Step 4: Live pushback (iterative - the heart of the skill)

Not one shot. Rehearse the room.

1. Enter the persona fully - adopt the tone from the atlas. Announce it briefly: [As your skeptical CFO].
2. Round 1: fire the 3-4 sharpest questions, calibrated to persona x focus area, weighted toward the weaknesses found in Step 3. Pull from the challenge banks below. Do not dump 10 - a real critic opens with the ones that hurt.
3. User answers. Then react as the real persona would:
   • Thin/evasive answer → press harder, follow-up on the same thread (don't let it go).
   • Solid answer → acknowledge briefly, move to the next weakness or dimension.
   • When an answer reveals a deeper assumption → escalate with a method (Step 5).
4. Continue 2-4 rounds until the sharp material is spent or the user calls stop. Stay in character; never fabricate the user's facts.
5. Offer to switch persona ("now let legal take a pass") or go to the summary.

Challenge banks by dimension (killer questions - baked in)

Draw from the dimensions relevant to persona x focus. These are the highest-severity ("killer") probes; expand naturally in-character.

• Financial / unit economics: "What's your monthly burn and runway?" · "What's CAC and payback period?" · "How did you compute LTV - what if it's far lower?" · "What's the margin per unit after all costs?" · "When and how do you reach profitability?"
• Market size & demand: "What's your TAM and how did you calculate it?" · "What proof that customers need this and will pay?" · "What if the market grows at 5%, not 20%?" · "What if a big incumbent enters before you get position?"
• Competition & differentiation: "Who are your real direct + indirect competitors and what do they do better?" · "Why can't Google/Amazon own this in a few months?" · "If your know-how isn't protected, how long to copy you?"
• Technical feasibility & scale: "Do you have a working prototype - proven in real conditions?" · "How does the architecture handle 10x load?" · "Which core tech is unproven, and what if it fails?" · "Key-person risk - what if your lead engineer leaves?"
• Execution / resources / timeline: "What happens if a milestone slips - is there a plan B?" · "Who owns each critical task, and who's the backup?" · "What will you cut to hit the deadline?"
• Prioritization / opportunity cost: "Why is this the priority over other options - what do we lose by not doing something else?" · "Same capital in product B / market Y - what's the return?" · "What's in the MVP vs deferred, and why?"
• Risk / legal / security / privacy: "Which legal/regulatory approvals are needed - do you have them all?" · "How do you protect customer data (GDPR/HIPAA)?" · "Pen-test done? Incident plan if breached?" · "Who's liable and what's the insurance if it goes wrong?"
• Evidence & assumptions: "Which key assumptions have you actually tested in the field?" · "What's the single most critical assumption - what if it breaks?" · "On what data are your conversion/retention/price estimates based?"
• Strategic fit / why now: "Why is now the right moment - what changed vs last year?" · "If a competitor ships something similar in months, how do you react?" · "Proof early adopters feel urgency now, not just hype?"
• User desirability: "Why would customers pick you - what value?" · "Is this must-have or nice-to-have, and how do you drive adoption?" · "Usage data (retention, active users) from pilots?"
• GTM & distribution: "How fast and at what cost can you scale via these channels?" · "What if one channel dies - what's the backup?" · "Biggest risk in your sales funnel?"
• Moat / defensibility: "What real barriers protect you (patents, network effects, switching cost)?" · "How long for a funded competitor to reach your market?" · "If the business model collapses, what value remains?"
• Scope / MVP discipline: "What MUST the MVP contain, and what goes to v2?" · "Name one feature you added later" (probes creep) · "If you had to halve scope, what gets cut first?"

Step 5: Method escalation (optional, by stakes)

When an answer exposes a deeper issue, or the decision is high-stakes, run a structured method instead of more scattered questions.

Method combinations by stakeholder: investors/board → pre-mortem + red team + outside view; product team → 5 Whys + riskiest-assumption + "so what/why now/why you"; strategy → falsification + outside view + second-order.

Real-room reference (what each room judges you against)

When the persona maps to a formal forum, hold the answer to that room's bar and name its top rejection reasons in the summary.

Step 6: Prep summary (the deliverable)

Always end here - this is what the user takes to the real room.

# Stress Test - [Proposal name]

**Challenged as:** [persona(s)]
**Focus:** [area(s)]
**Method(s) used:** [if any]
**Date:** [YYYY-MM-DD]

## Held up
[Answers/points that survived pressure - the user can lean on these.]
- [point] - [why it held]

## Thin
[Weak or evasive answers - where the user struggled.]
- [point] - [why it's thin] - [the exposing question they couldn't fully answer]

## Unresolved blind spots
[Real gaps that surfaced and are NOT yet answered - the danger list.]
- [gap] - [which persona pounces on it]

## Robustness score
[Score the proposal on the dimensions that matter for this room - weak / adequate / strong per dimension, from the rubric.]

| Dimension | Rating | Note |
|---|---|---|
| Problem & need | [weak/adequate/strong] | [one line] |
| Solution / USP | | |
| Market | | |
| Competition | | |
| Traction / demand | | |
| Unit economics | | |
| Execution & focus | | |
| Timeline & goals | | |
| Risk & kill criteria | | |

## Pre-meeting checklist
[Concrete, prioritized actions before the real room.]
- [ ] [prepare X / strengthen Y / cut Z / get data on W]

Route the blind spots (offer, don't impose): any item in "Unresolved blind spots" that is an untested assumption belongs in a register, not just in this prep file. Offer to append them: validation-stage assumptions → Hypothesis Register (/pm-hypotheses), project-level risks → Assumptions & Risks Register (from /pm-project-charter). A blind spot that only lives in meeting-prep notes gets rediscovered by the next skeptic; one in a register gets tested.

Internal completeness checklist

• Persona resolved (given, or via AskUserQuestion - never silently guessed)
• Focus resolved (given or inferred-and-stated)
• Silent weakness diagnosis run before role-play
• Live mode: multiple rounds, in-character, pressed on thin answers (not one shot)
• Never fabricated the user's facts - challenged gaps/assumptions instead
• Summary: held / thin / blind spots / robustness score / pre-meeting checklist

Save to

pureinn-workspace/[project-slug]/stress-tests/[YYYY-MM-DD]-[persona]-[topic-slug].md

stress-tests/ is a cross-cutting operational folder (like meetings/, prototypes/) - created on demand, not part of the phase artifact flow. Save only if the user wants a record; a quick rehearsal can stay in-chat.

Handoff

Čo si teraz má: Prešiel si si najostrejšie otázky, ktoré ti reálny stakeholder položí - vieš, čo obstálo, kde si tenký, a čo máš pred stretnutím dopripraviť. Menej prekvapení v reálnej miestnosti.

Ďalší krok:

• Slabé miesta v dátach → skill, ktorý ich doplní (/pm-market-analysis, /pm-business-case, /pm-hypotheses, /pm-personas).
• Ideš pitchovať → /pm-pitch-deck s posilnenými odpoveďami.
• Chceš druhý uhol → spusti /pm-stress-test znova s inou personou (napr. po CFO ešte legal).

Môžeš preskočiť ak: Rozhodnutie je nízko-rizikové, reverzibilné a nikto ho reálne nebude challengovať - vtedy je rehearsal zbytočná réžia.