Skip to content

Repository files navigation

Claude-Code (Ruflo Workspace)

Personal monorepo and Claude Code workspace for Henry Goldsmith. Contains multiple production-ready web apps, automation scripts, a self-improving knowledge base, and extensive Claude agent skills/commands.

Structure

Path Purpose
apps/ Standalone Next.js / Vite / Python / Electron applications
.claude/ Claude Code skills, agents, commands, helpers, settings
src/workspace_daemon/ Background workspace daemon (Google/MSFT integration)
execution/ One-off Python automation scripts (leads, email, invoices, etc.)
raw/ Append-only knowledge ingestion zone
wiki/ Indexed knowledge base (read this first)
output/ Improvement-loop artifacts and review files
CLAUDE.md Project rules, agent coordination, swarm config

Apps

Source of truth: apps/registry.json (machine-readable, validated by scripts/validate-registry.mjs). Deployment source of truth: docs/deployment-dependency-map.md, enforced by each app's Vercel ignoreCommand. Lifecycle: active = shipped, incubating = in dev, maintenance = bugfix-only, archived = frozen reference, superseded = replaced, tooling/service = internal, external = standalone repo.

App Path Stack What it is
Arise apps/arise Vite + React + Tailwind Training & progression (programs, sessions, levelled attributes)
Daily Debate apps/daily-debate Next.js + Supabase + Anthropic Daily critical-thinking debates (solo vs AI + PvP) with argument graph
Reflect apps/emotion-tracker · site apps/reflect-site Next.js + Anthropic Structured reflection event→assumptions→alternatives→action→follow-up with hedged bias language + longitudinal calibration
Forq apps/food-shopping-os · site apps/forq-site Next.js (local-first) Food shopping & planning OS (pantry, retailer, nutrition)
Le Studio French apps/french-practice · hub apps/le-studio-site Vite + React + Tailwind French PWA (Today/Speak/Review/Learn/Progress, Groq arena) — family hub is Le Studio/Forq/Chrono
Habit apps/habit-tracker Next.js + Supabase Quiet habit tracker with daily check-ins, targets and Pulse history
Revise apps/revise Next.js + Supabase + FSRS GCSE + A-level revision across WJEC/AQA/Edexcel/OCR (2,216 specPoints, validator)
Rapport apps/rapport Next.js (local-first) + optional Supabase Adaptive social-skills training — skill graph, mastery model, conversation simulator, real-world challenges (works with no AI provider)
Noticed apps/mental-load-tracker Next.js + Supabase Realtime Shared board for household invisible labor (two-person, no leaderboard)
rtk apps/rtk Node CLI Token-saving filter for noisy tool output (per-tool parsers + redaction)

Also in this repo (not product apps, but tracked so the registry doesn't drift): packages/le-studio-tokens (shared design tokens), src/workspace_daemon (background sync service), .claude (skills/agents/helpers incl. prompt-improver), vendor/life-os-scrape (archived — frozen Life OS build, superseded by arise; see its README). External: Chrono (henrygoldsmith07-wq/chrono-calendar, ICS/JSON interop).

Engineering standards

Document Purpose
docs/definition-of-done.md Minimum quality gates per product (install, lint, type-check, tests, build, E2E, security, perf)
docs/ci-coverage-audit.md Which workflow runs which gate for which app; live gap list
docs/ai-standard.md Shared AI pipeline: validation → context → provider → schema → deterministic gates → confidence → provenance → output
docs/security-standard.md Auth/RLS/tenancy/secrets/SSRF/rate-limit audit + enforcement
docs/observability.md · packages/observability/ Privacy-safe AI telemetry & error records (no user content by construction)
docs/data-provenance.md Source/timestamp/freshness/transformation/confidence/version for derived insights
docs/migration-backup.md Migration re-runnability, RLS re-verification, backup/export/recovery per DB-backed app
docs/a11y-baseline.md · docs/a11y.md Accessibility baseline + Le Studio family checklist
docs/perf-budgets.md · scripts/performance-budgets.json Bundle budgets enforced in CI
docs/evidence-registry.md · evidence/registry.json Durable claim ledger: demonstratedexternally validated, never auto-upgraded
docs/validation-state.md Per product: implemented vs proven effective — kept strictly separate

Knowledge Base & Self-Improvement

See CLAUDE.md for the full contract. Key commands:

  • /data-ingestion — Sync sessions, ecosystem data, curated content → raw/ + wiki/
  • /improve-system — Auto-apply low-risk hygiene; propose higher-stakes changes for sign-off

Routines run Tue/Fri (UK time) on the knowledge-base-updates branch.

Quick Start (Claude Code)

claude mcp add claude-flow -- npx -y ruflo@latest mcp start
npx ruflo@latest doctor --fix

License

MIT — see LICENSE.

Releases

Packages

Contributors

Languages