---
project:
  slug: "my-project"
  status: "in_progress"
  title: "My project"
  tagline: "One line about what it does."
  family: "Other / Various"
  demoUrl: ""
  aiModels: ["Model A", "Model B"]
  ideOrAgent: "Kilo"
  authorName: "Your name"
  hardwareStack: ["Host / hardware used"]
  reqHw: ["Any Docker host or VPS", "1 vCPU", "1 GB RAM", "10 GB free disk"]
  reqSw: ["Docker + docker compose", "runtime (Python/Node)", "VS Code + AI Coding Agent (Claude Code, Kilo, Cline..) or Desktop AI Clients"]
  reqAi: ["AI provider account, pay-per-token", "API key (OpenAI-compatible)"]
  totalCostUsd: 0.0
  timeToBuildHours: 1
  buildMinutes: 0
  buildDays: 1
  rulesFile: "docs/rules.md"
  description: |
    INITIAL GOAL
    What you set out to build.

    HOW IT WAS ORCHESTRATED
    Which agent + models, and how the work was split.

    REPRODUCE IT
    Point out the rules box; note the setup is tool-agnostic.

    OUTCOME
    What works now, and the token spend.
---

# <Project> — Project Journal

Milestone log (not provisional tweaks). Newest first. Timestamps in UTC.

---

## 2026-01-01 — Start

### M1 — 2026-01-01T00:00Z — First milestone

## Token & cost tracking
- Each milestone adds a cumulative `**Usage**` line (input/output/reasoning/cache/cost USD).
- Local source: `npm run usage` (Kilo DB).

## How to maintain
- New `## YYYY-MM-DD` entry at top for every closed milestone.
- Milestone header: `### M<n> — <UTC timestamp> — Title` (M1, M2, ...).
- One short line per step, sections for major milestones; skip minor tweaks.