nightytidy 0.3.7 → 0.3.9

package/src/prompts/steps/31-product-polish-ux-friction.md

@@ -1,122 +1,122 @@
-# Product Polish & UX Friction Audit
-
-READ-ONLY analysis. Do not modify any code.
-
-## Ground Rules
-
-- Evaluate as a **user**, not a developer. "The code handles this correctly" is irrelevant if the user can't tell.
-- Be specific: not "onboarding could be better" but "after signup, user lands on an empty dashboard with no guidance."
-- Classify every issue: **broken** / **confusing** / **incomplete** / **missing**.
-- Severity = frequency × pain. Trace every flow.
-
----
-
-## Phase 1: User Journey Mapping
-
-**Entry points** — Trace each: signup, login, invite link, OAuth, magic link, public pages, shared links, API, CLI, deep links.
-
-**Core journeys per user role:**
-- First use: signup → onboarding → first meaningful action → "aha moment"
-- Core loop: the daily/weekly workflow
-- Configuration: settings, profile, team/org management
-- Edge cases: account recovery, plan changes, data export, deletion
-- Exit points: dead ends, confusing branches, flows that just stop
-
-**Secondary flows** — Notifications, search, filtering, sorting, bulk actions, imports/exports, integrations, billing, admin.
-
----
-
-## Phase 2: First-Use & Onboarding
-
-**Signup:** Step count, field necessity, email verification clarity (cross-device?), OAuth permission scope & failure fallback, error specificity (duplicate email, weak password, etc.).
-
-**First experience:** What appears post-signup — empty state, tutorial, or sample data? Clear path to first action? Blocking setup steps? Skippable onboarding? Progress saved if user leaves?
-
-**Empty states:** For every list/dashboard/feed — what shows with zero data? Does it guide the user toward populating it?
-
----
-
-## Phase 3: Core Workflow
-
-**Primary workflow:** Click/step count for common actions. Unnecessary confirmations? Missing confirmations on destructive actions? Undo support? Save clarity (auto vs. manual, feedback)?
-
-**Forms & inputs:** Required/optional marking, inline vs. submit-only validation, sensible defaults, helpful placeholders, error display (all vs. first), input preservation on failure, progress for long forms, timezone/date format clarity.
-
-**Navigation:** Location awareness (breadcrumbs, active states, titles), back-navigation (browser + in-app), information architecture logic, deep link shareability & permissions.
-
-**Feedback & loading:** Immediate feedback on every action? Click-and-nothing-happens cases? Progress for long operations? Safe to navigate away? Retry without re-entry on failure?
-
----
-
-## Phase 4: Edge Cases & Errors
-
-**Destructive actions:** Confirmation with consequences explained? Undo available & obvious? Cascade effects communicated? Bulk action extra confirmation with count?
-
-**Common error states:** Network offline, session expired (unsaved work?), permission denied (actionable message?), not found (helpful or generic 404?), rate limited (wait guidance?), file upload failures (size/type/network — all communicated?).
-
-**Concurrency:** Two users editing same resource — conflict handling? Multi-tab state sync? Stale data refresh?
-
-**Boundaries:** Long text (truncation/overflow/layout break?), special characters/emoji/RTL, large datasets (1000+ items — pagination/virtualization/performance?), minimum-input functionality.
-
----
-
-## Phase 5: Settings & Configuration
-
-**Every setting:** Discoverable? Explained? Immediate or requires save? Resettable to default? Dangerous settings guarded?
-
-**Missing settings users would expect:** Notification preferences, display prefs, timezone, language, default views, keyboard shortcuts, data export.
-
-**Account management:** Change email/password/name? Delete account (clear, complete process)? Team invite/role/removal flows? Data fate on leave/deletion?
-
----
-
-## Phase 6: Notifications
-
-**Inventory all** emails, in-app, push, webhooks: trigger, content quality, user control (opt-out, frequency, channel).
-
-**Transactional:** Welcome email usefulness, password reset clarity & expiry, invite context, billing transparency.
-
----
-
-## Phase 7: Accessibility Quick Scan
-
-Flag obvious issues only (defer full audit): keyboard-only core flow completion, color-only information, screen reader labels on interactive elements, mobile responsiveness.
-
----
-
-## Output
-
-Save as `audit-reports/31_PRODUCT_POLISH_REPORT_[run-number]_[date]_[time in user's local time].md`.
-
-### Report Sections
-
-1. **Executive Summary** — Overall polish level (rough/fair/good/polished), worst friction, journey health.
-2. **User Journey Map** — All flows traced, health per flow (smooth / some friction / significant friction / broken).
-3. **Critical Friction Points** — Table: Flow | Location (file/component) | Issue | Severity | Type
-4. **First-Use & Onboarding** — Signup friction, onboarding gaps, empty states.
-5. **Core Workflow** — Step-by-step assessment, friction, feedback, form quality.
-6. **Edge Cases & Errors** — Destructive action safety, error quality, boundaries.
-7. **Settings & Account** — Gaps, account management, configuration polish.
-8. **Notifications** — Inventory, quality, missing notifications, user control.
-9. **Accessibility Notes** — Obvious issues only.
-10. **Recommendations** — Priority-ordered by effort: quick fixes (hours) / medium (days) / larger (weeks).
-
-**Report rules:** Don't pad — if a flow is smooth, say so in one line. Note items requiring a running app as "verify in running app."
-
----
-
-## Chat Summary (Required)
-
-Print directly in conversation — don't make the user open the file.
-
-1. **Status Line** — One sentence: what you did.
-2. **Key Findings** — Most important friction points, specific and actionable.
-3. **Recommendations** (only if warranted):
-
-| # | Recommendation | Impact | Risk if Ignored | Worth Doing? | Details |
-|---|---|---|---|---|---|
-| *#* | *≤10 words* | *What improves* | *Low/Med/High/Critical* | *Yes/Probably/Only if time* | *1–3 sentences* |
-
-Order by risk descending. Be honest in "Worth Doing?" — if marginal, say so.
-
-4. **Report Location** — Full file path.
+# Product Polish & UX Friction Audit
+
+READ-ONLY analysis. Do not modify any code.
+
+## Ground Rules
+
+- Evaluate as a **user**, not a developer. "The code handles this correctly" is irrelevant if the user can't tell.
+- Be specific: not "onboarding could be better" but "after signup, user lands on an empty dashboard with no guidance."
+- Classify every issue: **broken** / **confusing** / **incomplete** / **missing**.
+- Severity = frequency × pain. Trace every flow.
+
+---
+
+## Phase 1: User Journey Mapping
+
+**Entry points** — Trace each: signup, login, invite link, OAuth, magic link, public pages, shared links, API, CLI, deep links.
+
+**Core journeys per user role:**
+- First use: signup → onboarding → first meaningful action → "aha moment"
+- Core loop: the daily/weekly workflow
+- Configuration: settings, profile, team/org management
+- Edge cases: account recovery, plan changes, data export, deletion
+- Exit points: dead ends, confusing branches, flows that just stop
+
+**Secondary flows** — Notifications, search, filtering, sorting, bulk actions, imports/exports, integrations, billing, admin.
+
+---
+
+## Phase 2: First-Use & Onboarding
+
+**Signup:** Step count, field necessity, email verification clarity (cross-device?), OAuth permission scope & failure fallback, error specificity (duplicate email, weak password, etc.).
+
+**First experience:** What appears post-signup — empty state, tutorial, or sample data? Clear path to first action? Blocking setup steps? Skippable onboarding? Progress saved if user leaves?
+
+**Empty states:** For every list/dashboard/feed — what shows with zero data? Does it guide the user toward populating it?
+
+---
+
+## Phase 3: Core Workflow
+
+**Primary workflow:** Click/step count for common actions. Unnecessary confirmations? Missing confirmations on destructive actions? Undo support? Save clarity (auto vs. manual, feedback)?
+
+**Forms & inputs:** Required/optional marking, inline vs. submit-only validation, sensible defaults, helpful placeholders, error display (all vs. first), input preservation on failure, progress for long forms, timezone/date format clarity.
+
+**Navigation:** Location awareness (breadcrumbs, active states, titles), back-navigation (browser + in-app), information architecture logic, deep link shareability & permissions.
+
+**Feedback & loading:** Immediate feedback on every action? Click-and-nothing-happens cases? Progress for long operations? Safe to navigate away? Retry without re-entry on failure?
+
+---
+
+## Phase 4: Edge Cases & Errors
+
+**Destructive actions:** Confirmation with consequences explained? Undo available & obvious? Cascade effects communicated? Bulk action extra confirmation with count?
+
+**Common error states:** Network offline, session expired (unsaved work?), permission denied (actionable message?), not found (helpful or generic 404?), rate limited (wait guidance?), file upload failures (size/type/network — all communicated?).
+
+**Concurrency:** Two users editing same resource — conflict handling? Multi-tab state sync? Stale data refresh?
+
+**Boundaries:** Long text (truncation/overflow/layout break?), special characters/emoji/RTL, large datasets (1000+ items — pagination/virtualization/performance?), minimum-input functionality.
+
+---
+
+## Phase 5: Settings & Configuration
+
+**Every setting:** Discoverable? Explained? Immediate or requires save? Resettable to default? Dangerous settings guarded?
+
+**Missing settings users would expect:** Notification preferences, display prefs, timezone, language, default views, keyboard shortcuts, data export.
+
+**Account management:** Change email/password/name? Delete account (clear, complete process)? Team invite/role/removal flows? Data fate on leave/deletion?
+
+---
+
+## Phase 6: Notifications
+
+**Inventory all** emails, in-app, push, webhooks: trigger, content quality, user control (opt-out, frequency, channel).
+
+**Transactional:** Welcome email usefulness, password reset clarity & expiry, invite context, billing transparency.
+
+---
+
+## Phase 7: Accessibility Quick Scan
+
+Flag obvious issues only (defer full audit): keyboard-only core flow completion, color-only information, screen reader labels on interactive elements, mobile responsiveness.
+
+---
+
+## Output
+
+Save as `audit-reports/31_PRODUCT_POLISH_REPORT_[run-number]_[date]_[time in user's local time].md`.
+
+### Report Sections
+
+1. **Executive Summary** — Overall polish level (rough/fair/good/polished), worst friction, journey health.
+2. **User Journey Map** — All flows traced, health per flow (smooth / some friction / significant friction / broken).
+3. **Critical Friction Points** — Table: Flow | Location (file/component) | Issue | Severity | Type
+4. **First-Use & Onboarding** — Signup friction, onboarding gaps, empty states.
+5. **Core Workflow** — Step-by-step assessment, friction, feedback, form quality.
+6. **Edge Cases & Errors** — Destructive action safety, error quality, boundaries.
+7. **Settings & Account** — Gaps, account management, configuration polish.
+8. **Notifications** — Inventory, quality, missing notifications, user control.
+9. **Accessibility Notes** — Obvious issues only.
+10. **Recommendations** — Priority-ordered by effort: quick fixes (hours) / medium (days) / larger (weeks).
+
+**Report rules:** Don't pad — if a flow is smooth, say so in one line. Note items requiring a running app as "verify in running app."
+
+---
+
+## Chat Summary (Required)
+
+Print directly in conversation — don't make the user open the file.
+
+1. **Status Line** — One sentence: what you did.
+2. **Key Findings** — Most important friction points, specific and actionable.
+3. **Recommendations** (only if warranted):
+
+| # | Recommendation | Impact | Risk if Ignored | Worth Doing? | Details |
+|---|---|---|---|---|---|
+| *#* | *≤10 words* | *What improves* | *Low/Med/High/Critical* | *Yes/Probably/Only if time* | *1–3 sentences* |
+
+Order by risk descending. Be honest in "Worth Doing?" — if marginal, say so.
+
+4. **Report Location** — Full file path.

package/src/prompts/steps/32-feature-discovery-opportunity.md

@@ -1,128 +1,128 @@
-# Feature Discovery & Opportunity Audit
-
-Read the entire codebase. Identify features, capabilities, and improvements worth building — grounded purely in what exists, what's partial, and what the architecture supports.
-
-**READ-ONLY. No web search. No code changes.**
-
----
-
-## Rules
-
-- Every recommendation must reference specific files, models, or patterns.
-- Distinguish: **natural extensions** (80%+ done), **logical additions** (users would expect), **ambitious opportunities** (differentiators).
-- Quality over quantity. 10 well-reasoned opportunities > 50 shallow ones.
-- Be honest about effort and maintenance burden. "Add AI" is not a recommendation — specify data, infrastructure, and minimal viable version.
-- Don't recommend features that conflict with the product's design intent.
-- Prioritize features leveraging existing data/infrastructure over new systems.
-- You have all night. Read everything.
-
----
-
-## Phase 1: Deep Codebase Understanding
-
-**Product model** — What it does, who it serves, every feature, the full data model (entities, relationships, collected data), user roles/permissions, monetization (free/paid/tiers/gating), integrations.
-
-**Architecture capabilities** — Background jobs, notification systems (email/push/in-app/webhooks), file handling, search (full-text/filtering/faceting), real-time (WebSockets/SSE), API surface & patterns, event/audit tracking.
-
----
-
-## Phase 2: Unfinished & Abandoned Features
-
-**Partially built features** — Look for:
-- DB tables/columns with no UI or API exposure
-- Models/types defined but unused in routes/components
-- Feature flags permanently off (read the guarded code)
-- Routes/endpoints not linked from UI; unreachable components/pages
-- TODO/FIXME comments describing planned features
-- Migrations adding schema for unfinished features
-- Config/env vars for unintegrated services
-
-For each: what was it, how far did it get, what would finish it?
-
-**Vestigial infrastructure** — Libraries barely used, notification infra sending only one type, permission systems more granular than needed, underutilized search/webhook/queue systems. These are sunk investment awaiting ROI.
-
----
-
-## Phase 3: Data-Driven Opportunities
-
-**Inventory all collected data** — User actions/events, timestamps, entity relationships, stored-but-unsurfaced metadata, computed-but-undisplayed aggregations.
-
-**Underutilized data** — Analytics/insights, personalization signals, automation triggers, collaborative signals, historical trends. For each: what data exists → what feature it enables → existing pipeline support → effort.
-
-**Missing data** — Features that need data not yet collected. What's the minimal collection that unlocks the most value?
-
----
-
-## Phase 4: Pattern-Based Feature Discovery
-
-**Generalization** — Hardcoded reports → report builder. Single notification type → configurable system. Fixed workflow → customizable engine. Single integration → framework. Manual admin → self-service. Single export → multi-format. Fixed views → customizable dashboards.
-
-**Cross-entity features** — Unified search, activity feeds, bulk operations, broad tagging/categorization, universal comments/notes, import/export gaps.
-
-**Power user features** — Keyboard shortcuts, saved filters/views, bulk editing, templates, API access, advanced search, custom fields, scheduled/recurring actions.
-
-**Admin & ops** — Missing admin views, audit logging gaps, user impersonation, data export, usage analytics, health dashboards.
-
----
-
-## Phase 5: Automation & Intelligence
-
-**Automate manual processes** — Repetitive action patterns (macros), predictable status transitions, inferable data entry, condition-triggered notifications, manual cleanup tasks.
-
-**Smart defaults** — Fields users fill identically, likely next actions, adaptive settings, context-based pre-population.
-
-**AI-augmentable features** — Text generation/summarization, manual classification, semantic search, auto-tagging, NL summaries of data, answering questions from product data. For each: what's augmented, what data feeds it, what infra exists, minimal viable version.
-
----
-
-## Phase 6: Platform Opportunities
-
-**API-as-product** — Is the API exposable to third parties? What internal capabilities would externals pay for? Could webhook/event patterns power an integration ecosystem?
-
-**Multi-tenancy / white-label** — Tenant-aware data model? Configurable branding? Partner resale/embedding potential?
-
-**Extensibility** — Custom fields/views/workflows? Plugin architecture potential? Natural integration boundaries?
-
----
-
-## Output
-
-Save as `audit-reports/32_FEATURE_DISCOVERY_REPORT_[run-number]_[date]_[time in user's local time].md`.
-
-### Report Structure
-
-1. **Executive Summary** — Maturity assessment, opportunity count by category, top 5 highest-value, overall untapped potential.
-
-2. **Unfinished Features** — Table: Feature | Evidence (files/tables) | Completion % | Effort to Finish | Value | Recommendation
-
-3. **Underutilized Infrastructure** — Table: Infrastructure | Current Usage | Potential Usage | Effort | Value
-
-4. **Data Opportunities** — Underutilized: Data Available | Feature Enabled | Pipeline Support | Effort | Impact. Missing: Feature Desired | Data Needed | Collection Effort
-
-5. **Feature Opportunities** (main deliverable) — Per feature: Name/description, Category (natural extension / logical addition / ambitious), Evidence (specific code references), Existing foundation (% estimate), Effort (days/weeks/months with specifics), Impact, Dependencies, Priority (Critical / High / Medium / Nice-to-have)
-
-6. **Automation & Intelligence** — Manual→automated, smart defaults, AI opportunities with data/infra grounding.
-
-7. **Platform Opportunities** — API, multi-tenancy, extensibility assessments.
-
-8. **Recommended Build Order** — Priority sequence by dependencies and effort-to-value. Group: quick wins (days), medium (weeks), strategic (months).
-
----
-
-## Chat Summary (Required)
-
-Print directly in conversation — don't make the user open the report.
-
-1. **Status** — One sentence: what you did.
-2. **Key Findings** — Specific, grounded bullets. Lead with value. (e.g., "The `user_events` table tracks every action but nothing surfaces it — a dashboard is low-effort since `jobs/daily_stats.ts` already aggregates.")
-3. **Recommendations** table:
-
-| # | Recommendation | Impact | Risk if Ignored | Worth Doing? | Details |
-|---|---|---|---|---|---|
-| | ≤10 words | What improves | Low–Critical | Yes / Probably / Only if time | 1–3 sentences |
-
-Order by value descending. Be honest — not everything is worth the engineering time. If nothing worth building was found, say so.
-
-4. **Report Location** — Full path to the detailed report.
-
-If a phase yielded zero findings, say so in one line.
+# Feature Discovery & Opportunity Audit
+
+Read the entire codebase. Identify features, capabilities, and improvements worth building — grounded purely in what exists, what's partial, and what the architecture supports.
+
+**READ-ONLY. No web search. No code changes.**
+
+---
+
+## Rules
+
+- Every recommendation must reference specific files, models, or patterns.
+- Distinguish: **natural extensions** (80%+ done), **logical additions** (users would expect), **ambitious opportunities** (differentiators).
+- Quality over quantity. 10 well-reasoned opportunities > 50 shallow ones.
+- Be honest about effort and maintenance burden. "Add AI" is not a recommendation — specify data, infrastructure, and minimal viable version.
+- Don't recommend features that conflict with the product's design intent.
+- Prioritize features leveraging existing data/infrastructure over new systems.
+- You have all night. Read everything.
+
+---
+
+## Phase 1: Deep Codebase Understanding
+
+**Product model** — What it does, who it serves, every feature, the full data model (entities, relationships, collected data), user roles/permissions, monetization (free/paid/tiers/gating), integrations.
+
+**Architecture capabilities** — Background jobs, notification systems (email/push/in-app/webhooks), file handling, search (full-text/filtering/faceting), real-time (WebSockets/SSE), API surface & patterns, event/audit tracking.
+
+---
+
+## Phase 2: Unfinished & Abandoned Features
+
+**Partially built features** — Look for:
+- DB tables/columns with no UI or API exposure
+- Models/types defined but unused in routes/components
+- Feature flags permanently off (read the guarded code)
+- Routes/endpoints not linked from UI; unreachable components/pages
+- TODO/FIXME comments describing planned features
+- Migrations adding schema for unfinished features
+- Config/env vars for unintegrated services
+
+For each: what was it, how far did it get, what would finish it?
+
+**Vestigial infrastructure** — Libraries barely used, notification infra sending only one type, permission systems more granular than needed, underutilized search/webhook/queue systems. These are sunk investment awaiting ROI.
+
+---
+
+## Phase 3: Data-Driven Opportunities
+
+**Inventory all collected data** — User actions/events, timestamps, entity relationships, stored-but-unsurfaced metadata, computed-but-undisplayed aggregations.
+
+**Underutilized data** — Analytics/insights, personalization signals, automation triggers, collaborative signals, historical trends. For each: what data exists → what feature it enables → existing pipeline support → effort.
+
+**Missing data** — Features that need data not yet collected. What's the minimal collection that unlocks the most value?
+
+---
+
+## Phase 4: Pattern-Based Feature Discovery
+
+**Generalization** — Hardcoded reports → report builder. Single notification type → configurable system. Fixed workflow → customizable engine. Single integration → framework. Manual admin → self-service. Single export → multi-format. Fixed views → customizable dashboards.
+
+**Cross-entity features** — Unified search, activity feeds, bulk operations, broad tagging/categorization, universal comments/notes, import/export gaps.
+
+**Power user features** — Keyboard shortcuts, saved filters/views, bulk editing, templates, API access, advanced search, custom fields, scheduled/recurring actions.
+
+**Admin & ops** — Missing admin views, audit logging gaps, user impersonation, data export, usage analytics, health dashboards.
+
+---
+
+## Phase 5: Automation & Intelligence
+
+**Automate manual processes** — Repetitive action patterns (macros), predictable status transitions, inferable data entry, condition-triggered notifications, manual cleanup tasks.
+
+**Smart defaults** — Fields users fill identically, likely next actions, adaptive settings, context-based pre-population.
+
+**AI-augmentable features** — Text generation/summarization, manual classification, semantic search, auto-tagging, NL summaries of data, answering questions from product data. For each: what's augmented, what data feeds it, what infra exists, minimal viable version.
+
+---
+
+## Phase 6: Platform Opportunities
+
+**API-as-product** — Is the API exposable to third parties? What internal capabilities would externals pay for? Could webhook/event patterns power an integration ecosystem?
+
+**Multi-tenancy / white-label** — Tenant-aware data model? Configurable branding? Partner resale/embedding potential?
+
+**Extensibility** — Custom fields/views/workflows? Plugin architecture potential? Natural integration boundaries?
+
+---
+
+## Output
+
+Save as `audit-reports/32_FEATURE_DISCOVERY_REPORT_[run-number]_[date]_[time in user's local time].md`.
+
+### Report Structure
+
+1. **Executive Summary** — Maturity assessment, opportunity count by category, top 5 highest-value, overall untapped potential.
+
+2. **Unfinished Features** — Table: Feature | Evidence (files/tables) | Completion % | Effort to Finish | Value | Recommendation
+
+3. **Underutilized Infrastructure** — Table: Infrastructure | Current Usage | Potential Usage | Effort | Value
+
+4. **Data Opportunities** — Underutilized: Data Available | Feature Enabled | Pipeline Support | Effort | Impact. Missing: Feature Desired | Data Needed | Collection Effort
+
+5. **Feature Opportunities** (main deliverable) — Per feature: Name/description, Category (natural extension / logical addition / ambitious), Evidence (specific code references), Existing foundation (% estimate), Effort (days/weeks/months with specifics), Impact, Dependencies, Priority (Critical / High / Medium / Nice-to-have)
+
+6. **Automation & Intelligence** — Manual→automated, smart defaults, AI opportunities with data/infra grounding.
+
+7. **Platform Opportunities** — API, multi-tenancy, extensibility assessments.
+
+8. **Recommended Build Order** — Priority sequence by dependencies and effort-to-value. Group: quick wins (days), medium (weeks), strategic (months).
+
+---
+
+## Chat Summary (Required)
+
+Print directly in conversation — don't make the user open the report.
+
+1. **Status** — One sentence: what you did.
+2. **Key Findings** — Specific, grounded bullets. Lead with value. (e.g., "The `user_events` table tracks every action but nothing surfaces it — a dashboard is low-effort since `jobs/daily_stats.ts` already aggregates.")
+3. **Recommendations** table:
+
+| # | Recommendation | Impact | Risk if Ignored | Worth Doing? | Details |
+|---|---|---|---|---|---|
+| | ≤10 words | What improves | Low–Critical | Yes / Probably / Only if time | 1–3 sentences |
+
+Order by value descending. Be honest — not everything is worth the engineering time. If nothing worth building was found, say so.
+
+4. **Report Location** — Full path to the detailed report.
+
+If a phase yielded zero findings, say so in one line.