quests

name

quests

description

Track and guide humans through complex multi-step real-world processes. Use when a user needs help with a bureaucratic, legal, technical, or any multi-step procedure that requires organized tracking, step-by-step guidance, and progress monitoring. Triggers on requests like "help me with this process", "guide me through", "track this project", "create a quest", or any complex task that benefits from being broken into manageable steps presented one at a time. Also triggers for existing quests: "how's my quest going", "what's next on [process]", "update my progress". This is for multi-session/multi-day processes, not simple one-off tasks. The quest replaces scattered memory files — it IS the memory for long-running processes.

Quests — Guided Process Framework

A standardized framework for AI agents to track and guide humans through complex long-term tasks. The quest is the single source of truth — context, decisions, contacts, risks, and progress live inside the quest, not scattered across memory files.

Philosophy

• One step at a time: quest next shows only the current task — no overwhelm
• Quest as memory: quest context gives the agent everything needed in minimal tokens
• Living document: Steps can be added, removed, reordered, and modified at any time
• Human-friendly: quest brief generates summaries suitable for messaging

CLI: skills/quests/scripts/quest.py (symlink as quest)

Data stored at $WORKSPACE/data/quests.json. Quest IDs are auto-generated from names (slugified).

Conventions

• Auto-resolution: When only one quest is active, the quest argument is optional
• Fuzzy matching: Quests match by exact ID, ID prefix, or name substring