Skip to content
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
57 changes: 0 additions & 57 deletions .claude/settings.json

This file was deleted.

32 changes: 0 additions & 32 deletions .dev/notes..txt

This file was deleted.

81 changes: 81 additions & 0 deletions .dev/notes.md
Original file line number Diff line number Diff line change
@@ -0,0 +1,81 @@
## /gsd:discuss-phase — “vision decision” checklist

These phases need your input because the plans had to assume UX/presentation details.

### Discuss first (highest risk / most UX ambiguity)

- **Phase 3 — Onboarding Wizard** (`/gsd:discuss-phase 3`)
- **Why**: user-facing interactive CLI
- **Decide**: question flow, presentation style, validation behavior, help text, branching vs linear

- **Phase 8 — Rolling Tracker & Hedge Sizer** (`/gsd:discuss-phase 8`)
- **Why**: most complex phase (6 plans); CLI design heavy
- **Decide**: subcommand naming, output format (`status` / `suggest-roll` / `history`), how roll suggestions display

- **Phase 10 — Template Engine** (`/gsd:discuss-phase 10`)
- **Why**: visual interactive product
- **Decide**: explorer layout, node appearance, interaction patterns (click / hover / drag), color scheme

- **Phase 11 — Self-Assessment & Topics** (`/gsd:discuss-phase 11`)
- **Why**: interactive UX choices
- **Decide**: 4-state cycling behavior, guided vs standard vs YOLO differences, topic content depth

- **Phase 12 — Maya / CLI / Landing** (`/gsd:discuss-phase 12`)
- Landing page design + CLI ergonomics (templates exist; builds on phases 10–11)
- **Phase 9 — SQQQ vs Puts** (`/gsd:discuss-phase 9`)
- Mostly presentation (math/formulas fixed by research)
- **Phase 7 — Total Return** (`/gsd:discuss-phase 7`)
- Mainly CLI output formatting (narrow scope; follows existing patterns)

### Skip (infrastructure; minimal gray areas)

- **Phase 1 — Git Scrub**: tool-dictated (git-filter-repo, gitleaks)
- **Phase 2 — Setup Automation**: scripts/automation, minimal UX
- **Phase 4 — Polish & Hooks**: mostly technical (SIGINT, Bun migration)
- **Phase 5 — Agent Readiness**: needs planning, but scope is lint/templates/coverage (low UX ambiguity)
- **Phase 6 — Config & Models**: internal infrastructure for downstream CLIs

---

## PR #8 status

**Status:** Ready to merge (pending checks)

| Item | Status |
| --- | --- |
| Merge conflicts | Resolved — rebased clean branch with only Phase 11 commits |
| Mergeable | Yes |
| CodeRabbit | Re-reviewing (triggered by force push) |
| GitGuardian | In progress |
| Claude Review | Queued |

### Codex GPT-5.2 review findings

#### Good

- All prior CodeRabbit issues addressed
- Plan docs are well-structured with consistent frontmatter and verification blocks
- Persistence architecture is sound (per-topic keying, version-aware, migration-aware)

#### Issues to note (non-blocking; plan docs, not code)

- **"7 knowledge values" but only 6 listed** (low)
- Detail: Plan says 7, but the enum lists 6: `know`, `fuzzy`, `unknown`, `familiar`, `confident`, `mastered`. Fix during execution.
- **"Yolo" vs "YOLO" casing mismatch** (low)
- Detail: Context says "Yolo", plans use "YOLO". Cosmetic — resolve at implementation.
- **Color mapping inconsistency** (low)
- Detail: Context says gray/blue/green/gold; plans hardcode red/amber/blue/green. Plans are more specific; context is aspirational.
- **`src/explorer/` doesn’t exist on branch** (expected)
- Detail: Phase 10 creates those files. Plans reference future paths correctly.
- **localStorage fallback is export-only** (noted)
- Detail: No import path — acceptable for v1; can add in a later phase.

None of these block merge. They’re plan-level inconsistencies that get resolved during Phase 11 execution.

### Learn (7/7)

**Summary:** PR #8 merge conflicts resolved via branch rebase. Now mergeable, awaiting automated checks (CodeRabbit, GitGuardian, Claude Review). Codex review found no blocking issues.

**Next:** Wait for checks to pass, then merge. You can resume Codex with `codex resume` for follow-up analysis.

**Obi:** PR #8 is now mergeable. Checks are running — once they pass, it’s ready to merge. Codex found only cosmetic plan-doc inconsistencies, nothing blocking.
2 changes: 1 addition & 1 deletion .planning/phases/02-setup-automation/02-02-SUMMARY.md
Original file line number Diff line number Diff line change
Expand Up @@ -77,7 +77,7 @@ Each task was committed atomically:
3. **Task 3: Update integration test for new setup.sh behavior** - `1ef6a53` (test)

## Files Created/Modified
- `setup.sh` - Added 8 new functions: is_step_complete, mark_step_complete, show_progress, create_dir, create_directory_structure, verify_directory_structure, scaffold_file, scaffold_config_files, install_python_deps, print_summary. Updated main flow with step-based execution.
- `setup.sh` - Added 10 new functions: is_step_complete, mark_step_complete, show_progress, create_dir, create_directory_structure, verify_directory_structure, scaffold_file, scaffold_config_files, install_python_deps, print_summary. Updated main flow with step-based execution.
- `.gitignore` - Added .setup-progress exclusion in Family Office section
- `tests/integration/test_setup_onboarding_integration.sh` - Complete rewrite: 46 assertions across 5 test groups, removed all old onboarding/symlink/MCP references

Expand Down
Original file line number Diff line number Diff line change
@@ -0,0 +1,82 @@
# Phase 10: Template Engine & Dividend Topic Port - Context

**Gathered:** 2026-02-03
**Status:** Ready for planning

<domain>
## Phase Boundary

Build a pipeline that converts topic JSON + HTML template into standalone interactive knowledge explorers using Cytoscape.js, with the dividend strategy topic as the first output. Feature parity with the existing prototype (`.dev/playgrounds/dividend-strategy-explorer.html`). Knowledge state persistence, learning modes, additional topics, and Maya integration are Phases 11-12.

</domain>

<decisions>
## Implementation Decisions

### Graph visualization engine
- Use Cytoscape.js library instead of the prototype's custom canvas renderer
- Force-directed layout (Cytoscape's `cose` or similar) to match the prototype's organic feel
- Dark theme as default with a light mode toggle (both themes)
- Node colors based on category (Foundation=blue, Strategy=purple, etc.) — knowledge state shown via badge or border indicator, not node fill color
- Template defines a global color palette for category slots; topics assign category names and the template maps them to colors by order

### Explorer interaction model
- Clicking a node opens a bottom drawer detail panel (slides up from bottom, graph stays full width)
- Bottom drawer shows: concept description, related concepts, generated learning prompt, and a "Copy prompt" button
- Knowledge state cycling available via a button inside the detail panel (not by clicking the node directly)
- Preset category filter buttons that highlight/dim nodes by category group
- Text search box to find specific concepts by name (in addition to preset filters)
- Hovering a node highlights its connected edges and adjacent nodes (Cytoscape built-in)

### Topic data structure
- Full topic envelope: JSON includes topic title, description, version, author, categories array, concepts array, edges array
- Each concept: id, name, category, description, difficulty (beginner/intermediate/advanced), prerequisites (array of concept IDs)
- Edges use typed relationships: each edge has {source, target, type, label?} where type is one of (enables, requires, protects, funds, etc.) — type determines edge styling (color, dash pattern)
- Zod schema validates the entire topic JSON structure at build time
- Category colors defined in the template as a global palette, not in topic JSON — topics name their categories, template assigns colors by order

### Build pipeline output
- HTML + separate JS bundle per topic (not single-file inline)
- Cytoscape.js bundled into the JS output during build (works offline, no CDN dependency)
- Bun as the build tool (consistent with project's existing Bun usage)
- Strict Zod validation before build — malformed JSON fails with specific error messages, no broken HTML output
- Output directory: `fin-guru-private/explorers/` (private output, consistent with existing conventions)
- Build command produces: `{topic-slug}.html` + `{topic-slug}.js` per topic

### Claude's Discretion
- Exact Cytoscape.js layout parameters and animation settings
- Bottom drawer animation and sizing details
- Edge type-to-style mapping (which colors/dashes for each relationship type)
- Search implementation details (fuzzy vs exact matching)
- Bun build script internals and file watching for dev mode
- Dark/light theme color palettes and toggle mechanism

</decisions>

<specifics>
## Specific Ideas

- The existing prototype at `.dev/playgrounds/dividend-strategy-explorer.html` is the authoritative reference for feature parity — 21 concepts, 6 categories, 22 edges, same graph structure
- The prototype uses a GitHub-dark aesthetic — the dark theme should feel similar
- Generated learning prompts should be concept-specific and useful when pasted into an AI chatbot
- The detail panel should feel contextual but not obstructive — bottom drawer keeps the graph visible above it

</specifics>

<deferred>
## Deferred Ideas

- Knowledge state persistence via localStorage — Phase 11 (EXPL-05)
- Learning mode selector (guided/standard/yolo) — Phase 11 (EXPL-06)
- Options-greeks and risk-management topic content — Phase 11 (EXPL-09a, EXPL-09b)
- Mobile/touch polish — Phase 11 (EXPL-08)
- Maya learner profile export — Phase 12 (EXPL-07)
- Topic selector landing page — Phase 12 (EXPL-10)
- CLI launcher (`fin_guru.py explore`) — Phase 12 (EXPL-12)

</deferred>

---

*Phase: 10-template-engine-dividend-topic-port*
*Context gathered: 2026-02-03*
Loading