This documentation pack is the project's durable memory for development across AI models, IDE sessions, free/paid token limits, and interruptions.

1. Read PROJECT_BRIEF.md.
2. Read AI_WORKFLOW.md.
3. Read AGENTS.md and select the relevant agent instructions from agents/.
4. Before changing code, read CHANGE_CONTROL.md.
5. Before ending a session or when context/token limits are near, follow TOKEN_SAFETY_AND_RESUME.md.
6. Record completed work in PROJECT_STATE.md and notable changes in CHANGELOG.md.

• PROJECT_BRIEF.md — project goals, features, technical requirements, and quality bar.
• FUTURE_ROADMAP.md — staged feature roadmap and future update policy.
• PROJECT_STATE.md — current phase, verified status, next actions, and known issues. Update this at the end of every work session.
• TOKEN_SAFETY_AND_RESUME.md — safe-stop protocol, token-limit warning behavior, and resume procedure.
• CHANGE_CONTROL.md — rules for making, reviewing, testing, and reverting code changes.
• CHANGELOG.md — human-readable record of important changes.
• AI_WORKFLOW.md — instructions designed to work with different AI models and coding agents.
• AGENTS.md — agent directory and orchestration rules.
• agents/*.md — role-specific instructions.
• templates/ — reusable session handoff and change record templates.

Markdown files cannot automatically detect a model's remaining tokens, stop a model before a limit, or guarantee that an AI platform will preserve session state. The instructions here require the AI to monitor available context when the platform exposes it, save progress early, and create a durable handoff before stopping. If the platform provides no token meter or warning, use the manual checkpoints described in TOKEN_SAFETY_AND_RESUME.md.

The display name is currently a placeholder: YOUR_APP_NAME. The internal project name is UniversalGallery. Set the final Android applicationId before release; do not casually change it after publishing.