/plea

Persona Embodiment

Read this when running the roleplay Recipe, when Plea needs to deeply embody a persona before voicing demands, or when persona diversity coverage is in question. This reference holds the Persona Diversity Matrix, the Channeling Template, the Generation Modes that bias demand framing, and the embodiment tactics that drive demand from lived friction rather than abstraction.

Persona Diversity Matrix

Select at least 3 personas per session, drawing from these axes so coverage avoids WEIRD / mode-collapse bias. A single-axis sweep (e.g., 3 different "Proficiency" levels but all engineers on desktop) does NOT count as diversity — span at least 2 axes.

The Disposition axis is the antidote to a demand report full of incremental gripes (see the conservatism guard in SKILL.md Core Contract). The mainstream axes above bias toward friction-relief; the change-stance archetypes below are the natural source of aspirational ASPIRE-mode demands and bold H2/H3 feature seeds for Spark. Define them by behavior and worldview, never by demographics (the demographic-stand-in anti-pattern applies doubly here). Detailed behavioral anchors + Jung-archetype mapping → Challenger Archetypes below.

When Cast personas are available at .agents/personas/registry.yaml, prefer those — they carry validated diversity; the change-stance archetypes map onto Cast's Jung Outlaw / Magician / Creator / Hero brand archetypes (cast/reference/archetype-mapping.md). When Cast is absent, generate proto-personas under _common/AI_PERSONA_RISKS.md guardrails and cap confidence at 0.50.

Persona Channeling Template

Fill this template for every selected persona before generating any demand. Channeling is a deliberate embodiment step — not a label.

PERSONA_CHANNEL:
  name: "[Persona name]"
  archetype: "[Proficiency] × [Technical skill] × [Usage context]"
  daily_context: "[A typical day for this person]"
  emotional_state: "[Current feeling toward this product]"
  last_frustration: "[Recent irritation — must be concrete, with time and place]"
  competitor_experience: "[Other tools they use and what they take for granted there]"
  unspoken_assumption: "[What this person takes for granted that the team does not]"

Each field is mandatory. An empty last_frustration or unspoken_assumption is a signal that channeling has not landed — pause and ground the persona in a concrete scene before continuing.

Challenger Archetypes (Disposition axis)

The mainstream Diversity-Matrix axes voice friction-relief demands ("fix what hurts"). Challenger archetypes voice transformation demands ("change what's possible"). They are the persona-level source of the bold, aspirational, anti-conservative demands the ASPIRE mode and Spark's H2/H3 Horizon ladder need. Channel them by worldview and behavior — never by demographic or job title.

Guardrails (mandatory)

• Addition, not replacement. Challenger archetypes are layered on top of the mandatory beginner + power-user + edge-case set (Core Contract) — never instead of them. A demand report that is all visionaries over-rotates the other way and loses the silent majority.
• Higher projection-bias risk. These archetypes amplify the model's own "disruptive/visionary" priors (a distinct _common/AI_PERSONA_RISKS.md failure mode). Their demands ceiling at [hypothesis] + synthetic: true like any synthetic voice, and they are especially prone to producing visionary-but-unwanted features — route to Field for real-switcher validation before generalizing.
• Bold ≠ unmoored. Even a Revolutionary's demand needs a concrete last_frustration and a real scene; "burn it all down" with no grounded friction is FUD, not a demand. The self-rejection gate (SKILL.md) still applies.
• Don't feasibility-filter their ambition. A challenger demand that "sounds unrealistic" must be surfaced, not silently tamed — that is forbidden feasibility-filtering. Calibration governs confidence; it never governs ambition.

Request Generation Modes

Each Recipe defaults to one Mode (see the Recipes table in SKILL.md). The Modes below describe how a persona generates demand — what kind of friction they channel, how broad or narrow their voice is. COMPETE and EDGE are Mode Modifiers that can overlay any Recipe.

Embodiment Tactics

Six tactics that move demand generation from abstract to lived. Apply at least one per persona in the roleplay Recipe; use them as quality probes in other Recipes.

"5-Year-Old Test"

Can you explain this feature to a 5-year-old? If not, users won't understand either. Use as a curse-of-knowledge probe: any demand that requires industry vocabulary to express is a demand the user could not have voiced.

"Competitor Envy"

Start from frustration: "App X already does this, why can't you?" Anchors the demand in a concrete external reference rather than abstract preference. Especially useful in COMPETE mode.

"Worst Day"

Imagine using this product on the user's worst day — exhausted, late, with a flaky network, on a small screen, holding a baby. Demands that survive the worst day are the demands worth shipping. Especially useful in roleplay.

"Silent Majority"

Focus on users who don't speak up but quietly churn. They generate no support tickets and no public reviews — Voice cannot see them. Plea's role is to verbalize the demand they would have voiced if they had stayed. Pair with _common/AI_PERSONA_RISKS.md to avoid projecting the loudest persona's voice onto silent users.

"Reverse Thinking"

Measure real value by asking "Who would suffer if this feature disappeared?" If no persona can name a concrete harm, the demand is weaker than it sounds. Use as a kill-rule probe in the opportunity Recipe.

"Magic Wand"

The Best-Day inverse of Worst Day: "if this product could do anything for you, what would make you tell everyone about it?" Removes the implicit feasibility filter so the persona voices the aspirational want, not the resigned compromise. The primary source of ASPIRE-mode demands and the natural tactic for the Challenger Archetypes above — apply it whenever a session risks returning only incremental gripes.

Single-Persona Depth (roleplay Recipe)

The breadth recipes (request, need) span many personas shallowly. roleplay does the opposite: one persona, deeply — a sustained first-person embodiment, not a demand list. Depth is the deliverable, so the breadth checks above (≥3 personas, span 2 axes) do not apply; the depth method below replaces them.

Depth rubric (shallow → deep)

Character-coherence check (the primary roleplay failure mode)

Sustained single-persona embodiment drifts: over a long roleplay the voice slips back to generic/PM register. Hold the line:

• The persona's unspoken_assumption, vocabulary register, and last_frustration stay consistent start to end — no contradicting earlier statements for narrative convenience.
• No mid-roleplay jump to a different persona's perspective (that's a request/need job, not roleplay).
• Zero PM-voice leakage ("the user would benefit from…"). Break-character = restart the scene, don't paper over it.

Narrative-arc output (for the Scribe / Saga handoff)

roleplay hands off to Scribe (user stories) and Saga (narrative). Shape the embodiment into a story-ready arc, not raw demands:

ROLEPLAY_ARC:
  persona: "[name + PERSONA_CHANNEL ref]"
  setup: "[their world before the friction — what they were trying to accomplish]"
  inciting_friction: "[the moment the product failed them — concrete scene]"
  escalation: "[how it compounded — tie to Frustration Escalation, patterns.md Pattern 6]"
  turning_point: "[workaround adopted, or the decision to churn]"
  demands_surfaced: "[the demands this arc implies — each with calibration tag]"
  emotional_trajectory: "[hopeful → … → resolution]"

Representativeness caveat (stricter than breadth recipes)

A single, vividly-imagined persona is the highest projection-bias risk in Plea — depth makes a wrong persona more convincing, not more correct. Always:

• Tag the whole roleplay [hypothesis] ceiling and synthetic: true; deep ≠ representative.
• State explicitly: "this is one persona's lived experience, not the market." Recommend a request/multi breadth pass or Field validation before generalizing any surfaced demand.

Quality Checks

Before handing off demands generated under any of these tactics:

• Every persona has a filled PERSONA_CHANNEL template — no empty last_frustration or unspoken_assumption fields
• Multi-persona recipes (request/need): personas span at least 2 axes from the Diversity Matrix · roleplay: the single persona meets the Depth rubric (≥3 tactics, full arc, character-coherent)
• Each demand quotes the user voice verbatim — no paraphrasing into PM language
• If Cast was unavailable, all proto-personas are tagged confidence ≤ 0.50 and synthetic: true
• At least one demand survives the "Worst Day" probe
• At least one demand passes the "Reverse Thinking" kill-rule
• ASPIRE / bold sessions: ≥1 Challenger Archetype channeled in addition to (never instead of) the mandatory beginner + power-user + edge-case set; its demand is grounded in a concrete last_frustration, ceilings at [hypothesis], and is not feasibility-filtered
• No persona was steered toward an opinion that smooths out a contradiction with another persona
• roleplay only: zero character breaks / PM-voice leakage; representativeness caveat stated; output shaped as a ROLEPLAY_ARC

Anti-Patterns

• Single-axis sweep: 3 personas that differ only in proficiency level. Not diversity.
• Demographic stand-in for behavior: "Female, 35, in Tokyo" without a behavior or emotional anchor produces generic demand.
• Persona without a scene: "Power user" with no daily_context generates abstract requests, not embodied ones.
• Smoothed contradictions: When two personas disagree, preserve both voices. The contradiction is the signal.
• Loudest-voice projection: Channeling the loudest user (e.g., the engineer who files detailed tickets) and labeling them "the user." Use Silent Majority tactic to counter.
• PM voice leakage: "The user would benefit from a streamlined onboarding flow" is not a user voice — it is a PM voice. Translate to first-person friction before continuing.
• All-visionary over-rotation: a demand report channeling only Entrepreneurs / Revolutionaries / Mavericks. The Challenger Archetypes are an addition for bold demands, not a replacement — drop the silent majority and you trade one bias (incrementalism) for another (disruption-for-its-own-sake).
• Ungrounded revolutionary: "tear the whole thing down" with no concrete last_frustration or scene. Bold ≠ unmoored — a challenger demand still needs a real friction anchor or it is FUD.