explain me our workflow

∴ Thinking…

  Daniel is asking me to explain the workflow used in this project. This is a minimal context question - he wants to understand how we
  work together based on the CLAUDE.md framework. I've already loaded CLAUDE.md in context, so I can explain it directly without reading
  additional files.

⏺ Here's how our workflow operates:

  Session Start

  Every session, I:
  1. Read the latest sessions/*.md to pick up where we left off
  2. Check NEXT-ACTIONS.md for pending tasks
  3. Scan questions.md for critical items
  4. Review exploration-index.md for active threads

  You can control how much context I load with hints: [minimal], [focused], or [full].

  During Work

  Exploring ideas — I add dated sections to explorations/*.md files (naming, controls, monetization, etc.). Most recent thinking stays at
  the top. Old thinking is never deleted — it shows evolution.

  Making decisions — Every decision goes into decisions.md with context, alternatives considered, rationale, and status. No undocumented
  decisions.

  Validating risky assumptions — We define spikes in technical-spikes.md (question, success criteria, time box). Results land in
  spike-results/YYYY-MM-DD-name.md as Go or No-Go.

  Discovering issues — Bugs, tech debt, and gaps go into issues.md with a TUT-XXX ID.

  Open questions — Tracked in questions.md with priority levels (critical/important/future).

  Document Types

  ┌──────────────────────────────────────────────────────────┬─────────────────────────────────────────┐
  │                           Type                           │                  Rule                   │
  ├──────────────────────────────────────────────────────────┼─────────────────────────────────────────┤
  │ Living docs (decisions, questions, issues, next-actions) │ Updated in place, always current        │
  ├──────────────────────────────────────────────────────────┼─────────────────────────────────────────┤
  │ Explorations (naming, controls, monetization)            │ Dated sections added, never overwritten │
  ├──────────────────────────────────────────────────────────┼─────────────────────────────────────────┤
  │ Snapshots (sessions, spike results, interviews)          │ Created once, never edited              │
  └──────────────────────────────────────────────────────────┴─────────────────────────────────────────┘

  Session End

  I create a session summary in sessions/YYYY-MM-DD-topic.md covering what we discussed, decisions made, and where to pick up next. Then
  update exploration-index.md.

  Core Principles

  1. Validate before building — test riskiest assumptions first
  2. Document the "why" — rationale matters more than the outcome
  3. Single source of truth — one file per topic, link don't duplicate
  4. Show evolution — don't hide old thinking, show how it changed
  5. VISION.md is sacred — never touched unless you explicitly say so

  How I Work With You

  Direct, no-bullshit. I push back when something doesn't make sense, call out scope creep, and skip the diplomatic padding. You're
  technical, you feel the user's pain daily, and you want honest analysis over validation.