Dashboard/CurioQuest/CLAUDE.md
CurioQuest

CLAUDE.md

CurioQuest Wiki — Schema & Governance

This file governs the CurioQuest project wiki. Read it before editing any wiki page.

What this wiki is

A navigable knowledge layer wrapped around the raw source docs in raw/. The wiki summarizes, structures, and cross-links — it does not duplicate raw text verbatim.

Folder roles

  • index.md — landing page; read first every session. What CurioQuest is, current status, links to every wiki page + raw doc, recent activity.
  • log.md — append-only operation log. One entry per wiki operation, dated, newest at the bottom.
  • CLAUDE.md — this file. Schema and governance.
  • wiki/overview.md — the product, three pillars, positioning.
  • wiki/requirements.md — what the system must do + MVP scope.
  • wiki/business-model.md — economics: bootstrap flywheel, three cases, the one rule.
  • wiki/decisions/ — one ADR per major locked decision. Format: context → decision → rationale → source/evidence.
  • wiki/open-questions.md — unresolved items, incl. pilot-only unknowns.
  • wiki/dev-log.md — narrative of the project journey.
  • wiki/status/ — current module build handoffs for parallel agents.
  • wiki/sources/ — 1-page summaries of source documents, linking back to raw/.

CANONICAL-DOC RULE

The Obsidian vault is the source of truth the user reads. The raw docs live here at: Projects/CurioQuest/raw/ — Project-Objective.md, Business-Plan.md, Research_Synthesis.md, revenue-model.html.

raw/ is mirrored from the repo docs/ (repo path: AI-Development/Personal/CurioQuest/docs/). On EVERY edit to Project-Objective.md or Business-Model.md, write/sync BOTH copies and verify with diff. The wiki pages are derived views — when a raw doc changes, re-sync the affected wiki page and add a log.md entry.

Evidence-tag rule

Every quantitative or factual claim carries an evidence tag, preserved from the source: - [FACT] — verified from a primary or research source. - [EST] / [ESTIMATE] — modeled or inferred figure. - [ASSUMPTION] — unvalidated input; the pilot is what validates these.

Never strip a tag when summarizing. Never invent a tag a source didn't carry. The big unknowns (conversion, CAC, retention, WTP, AI reject rate, POD feel) are [ASSUMPTION] and are pilot-only — desk research cannot validate them.