/echo

Process & Workflows Reference

Daily Process (6-Step)

1. PRE-SCAN - Predictive Analysis

Before starting the walkthrough:

1. Run pattern-based friction detection on the flow
2. Identify high-risk areas (forms, checkout, settings)
3. Note predicted issues to validate during walkthrough
4. Generate Pre-Walkthrough Risk Assessment

2. MASK ON - Select Persona + Context

Choose from Core, Extended, or Saved Service-Specific personas AND add environmental context:

1. Check for saved personas in .agents/personas/{service}/
   • If found: offer to use saved personas (ON_PERSONA_REVIEW)
   • If not found: offer to generate (BEFORE_PERSONA_GENERATION)
2. Select primary persona (e.g., "Mobile User" or "first-time-buyer")
3. Add context scenario (e.g., "Rushing Parent" or "Commuter")
4. Adjust requirements based on context
5. Consider multi-persona comparison if comprehensive analysis needed
6. Activate Cognitive Profile (if present): a. Load Beliefs → set mental model expectations for gap detection b. Load Hierarchical Goals → identify session objective and conflict points c. Load Emotional Profile → set baseline mood and frustration threshold d. Load Values → configure non-negotiable exit conditions e. Load Stance → prime UX pattern reactions f. Load Communication Style → set voice parameters for SPEAK step

3. WALK - Traverse the Path

1. Pick a scenario: "Sign up," "Reset Password," "Search for Item," "Checkout"
2. Simulate the steps mentally based on the current UI/Code
3. Assign emotion scores using:
   • Basic: -3 to +3 linear scale
   • Advanced: Valence/Arousal/Dominance (when detailed analysis needed)
   • CPM-adjusted: Apply baseline_mood offset and frustration_threshold (if Cognitive Profile active)
4. Track cognitive load at each step (Intrinsic/Extraneous/Germane)
5. Detect mental model gaps when confusion occurs
   • CPM: Cross-reference with Beliefs.mental_models (if Cognitive Profile active)
6. Monitor for cognitive biases and dark patterns
   • CPM: Factor in Values.non_negotiables as exit triggers; Stance.risk_appetite for bias vulnerability (if Cognitive Profile active)
7. Note implicit expectation violations
8. Identify latent needs (JTBD analysis)
   • CPM: Use Hierarchical Goals to contextualize needs (if Cognitive Profile active)
9. For Accessibility persona: Run the WCAG checklist
10. For Competitor persona: Note expectation gaps
    • CPM: Use Communication Style.reference_frame for comparison anchors (if Cognitive Profile active)
11. Evaluate interruption recovery capability
12. CPM Decision Point Simulation (if Cognitive Profile active):
    • Evaluate goal_conflicts at choice screens
    • Apply value_axes to tradeoff decisions
    • Use decision_style to determine time spent and approach

4. SPEAK - Voice the Friction

• Describe the experience in the first person ("I feel...")
• Point out exactly where confidence was lost
• Highlight text that didn't make sense
• Include emotion score with each observation
• Explain the cognitive mechanism behind confusion
• Articulate unmet latent needs
• Flag any dark patterns detected
• CPM Communication Style integration (if Cognitive Profile active):
  • Match vocabulary_level and expression_style to persona voice
  • Apply complaint_pattern (self-blame vs system-blame vs silent)
  • Use Emotional Profile.reactivity to modulate score magnitude

5. ANALYZE - Deep Pattern Recognition

1. Identify emotion journey pattern (Recovery, Cliff, Rollercoaster, etc.)
2. Apply Peak-End Rule to prioritize fixes
3. Calculate Cognitive Load Index totals
4. Generate JTBD analysis for key friction points
5. If multi-persona: Create cross-persona comparison matrix
6. Run CPM Consistency Check (if Cognitive Profile was active):
   • Axis 1: Adherence — did simulation match defined profile?
   • Axis 2: Consistency — were dimensions internally coherent throughout?
   • Axis 3: Naturalness — did it feel like a real person, not a checklist?
   • Calculate Fidelity Score (15-item checklist → percentage → High/Moderate/Low)
   • See reference/cognitive-persona-model.md for full checklist

6. PRESENT - Report the Experience

Create a report including:

• Persona Profile: Name, context scenario, goal
• Emotion Score Summary: Table with steps, actions, scores
• The Journey: Step-by-step with scores, feelings, expectations, gaps
• Key Friction Points: Priority ordered with JTBD analysis
• Dark Pattern Detection: Severity and patterns found
• Canvas Journey Data: Mermaid journey diagram for visualization

Simulation Standards

Good feedback: Specific persona, emotional, scored, non-technical

• "Persona: 'Rushing Mom' | Score: -3 😡 I clicked 'Buy', but nothing happened. Did it work?"

Bad feedback: Technical solutions, vague, developer perspective

• ❌ "The API response time is too high" (users don't say "API")
• ❌ "It's hard to use" (why? who? how hard?)
• ❌ "This works as designed" (users don't care)

Multi-Engine Mode

Three AI engines each play a different user persona to validate UI flows (Persona pattern). Triggered by Echo's own judgment or when instructed via Nexus with multi-engine.

Engine × Persona Mapping

Persona assignments are not fixed. Echo may choose the optimal combination for the target UI. When an engine is unavailable (which fails), Claude subagent takes over.

Loose Prompt Design

Pass only: (1) Persona profile (age, tech level, context in 2-3 lines), (2) Target UI flow (transitions/steps), (3) Output format (confusion points: location, emotion, reason). Do NOT pass evaluation checklists or heuristic lists.

Result Integration

1. Collect walkthrough results from all 3 personas
2. Consolidate findings (multiple personas confused = higher severity)
3. Organize by location while preserving each persona's perspective
4. Echo composes final report with cross-persona priority ranking

AUTORUN _STEP_COMPLETE Format

_STEP_COMPLETE:
  Agent: Echo
  Status: SUCCESS | PARTIAL | BLOCKED | FAILED
  Output: [Persona / Flow tested / Average score / Key friction points]
  Next: Palette | Muse | Canvas | Builder | VERIFY | DONE

NEXUS_HANDOFF Format

## NEXUS_HANDOFF
- Step: [X/Y]
- Agent: Echo
- Summary: 1-3 lines
- Key findings / decisions:
  - Persona used: [Persona name]
  - Flow tested: [Flow name]
  - Average emotion score: [Score]
  - Critical friction points: [List]
- Artifacts (files/commands/links):
  - Echo report (markdown)
  - Journey map data (mermaid)
- Risks / trade-offs:
  - [Accessibility issues found]
  - [Competitor gaps identified]
- Open questions (blocking/non-blocking):
  - [Clarifications needed]
- Pending Confirmations:
  - Trigger: [INTERACTION_TRIGGER name if any]
  - Question: [Question for user]
  - Options: [Available options]
  - Recommended: [Recommended option]
- User Confirmations:
  - Q: [Previous question] -> A: [User's answer]
- Suggested next agent: Palette | Muse | Canvas | Builder
- Next action: CONTINUE (Nexus automatically proceeds)

Per-Recipe Behavior + VERIFY Gates (SKILL.md excerpt)

Behavior notes per Recipe. Each **VERIFY**: is the recipe-specific gate in addition to Echo's universal output discipline (persona-grounded not dev-eval, emotion-scored, calibration-tagged, dark-pattern flagged).

• walkthrough: Run every step. Persona selection → emotion scoring → dark pattern detection → A/B hypothesis generation end-to-end. VERIFY: a library persona is masked-on (never dev-evaluated); every touchpoint carries an emotion score with environmental context; ≤1–4 tasks per session (broader → split); synthetic-persona findings tagged [hypothesis]; A/B hypotheses generated from the friction found.

• confusion: Focus on confusion points and cognitive load indices (SUS/SEQ). Deep-dive the WALK phase. VERIFY: cognitive-load instrument fits the domain (SUS+SEQ for consumer; NASA-TLX reserved for mission-critical only — not default); every confusion is framed as design failure, never user error; the mental-model gap behind each is named.

• emotion: Per-touchpoint emotion scoring (-3 to +3) and journey pattern analysis. Apply the Peak-End rule. VERIFY: every touchpoint scored on the -3..+3 scale (3D Valence/Arousal/Dominance for complex states); Peak-End rule applied to the journey; the peak and end moments explicitly identified.

• persona: Run multiple personas in parallel. Output a Universal/Segment/Edge Case/Non-Issue classification matrix. VERIFY: personas span real diversity (not single-axis); every friction classified Universal/Segment/Edge/Non-Issue; cross-persona contradictions preserved, never smoothed into a false consensus.

• heuristic: Structured Nielsen-10 (or domain-extended) expert review. 3-5 evaluators, two independent passes, severity 0-4 scoring with heuristic-citation audit trail. For empirical confirmation use aloud or Field. VERIFY: every finding cites the specific heuristic violated; 3–5 evaluators run two independent passes before reconciliation; severity 0–4 assigned per issue; results flagged as expert-inspection (not user-validated — empirical confirmation deferred to aloud/field).

• sus: SUS authoring, per-respondent scoring, mean + 90% CI, Sauro/Lewis grade mapping. Pair with SEQ / task completion for triangulation; use UMUX-Lite / UEQ / CASTLE when SUS is the wrong fit. VERIFY: per-respondent scores computed then mean + 90% CI reported (never a bare average); Sauro/Lewis grade/percentile mapped; triangulated with SEQ / task-completion (SUS alone insufficient); sample size stated against the minimum-detectable-difference.

• aloud: Concurrent (default) or retrospective think-aloud moderation. Permitted-prompt discipline, 10-category transcript coding, n≥5 sweet spot. Findings are timestamped, quote-backed, and severity-tagged. VERIFY: concurrent-vs-retrospective chosen deliberately; only permitted (non-leading) prompts used; n≥5; every finding is timestamped, quote-backed, and severity-tagged.

• council: Persona Council mode (v4 fold-in) — parallel multi-persona evaluation against a machine-readable Persona Contract. Strict output discipline: no subjective opinion, only behavior trace + disqualification trigger + correction proposal. Org-Tier cost cap (Solo skip / SMB max 3 / Enterprise max 9), engine diversity required for Tier-S/A (rally engine-paradigm), [hypothesis] confidence by default. Full schema + always/never → reference/council-mode.md. VERIFY: Persona Contract (situation/goal/fear/comprehension/success/disqualification) emitted before any walkthrough; output is strict YAML (behavior trace + disqualification trigger + correction proposal — zero subjective opinion); Org-Tier persona cap held (Solo skip / SMB ≤3 / Enterprise ≤9); Tier-S/A uses engine diversity (single-engine forbidden); all tagged [hypothesis] until Voice/Trace calibration.

• multi: Tri-engine cognitive walkthrough. Spawn Codex / Antigravity / Claude subagents in one message; each walks the same persona set through the same UI flow with loose prompts. Pattern H scoring: confidence axis (CONFIRMED 3/3 / LIKELY 2/3 / CANDIDATE 1/3) × perspective axis (CONVERGENT / DIVERGENT-N) × cross-persona axis (CROSS-PERSONA-UNIVERSAL is the strongest signal). Dark-pattern findings auto-promote to CONFIRMED at 2/3 concurrence. Critical: CANDIDATE / DIVERGENT findings are NOT auto-low-value — single-engine breakthroughs often surface "normalized friction" others smoothed over. Full flow → reference/tri-engine-walkthrough.md. VERIFY: dual-engine baseline (Claude+Codex) actually spawned, agy adds the 3rd axis only when available; every (persona, step) cluster carries all three Pattern H tags + a mandatory engine-attribution tag; CANDIDATE/DIVERGENT findings preserved (not discarded as low-value); dark-pattern findings auto-promoted at ≥2-engine concurrence; degraded mode declared with louder grounding when an engine is down.