awf-session-restore

Restore AI session context lazily via /recap commands at three levels.

209|94|Updated Jan 15, 2026

Search Tags:#context #session #lazy-loading #recap #token-budget #AI-workflow

Ready to run

npx skills add https://github.com/TUAN130294/awf --skill awf-session-restore

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: awf-session-restore
Source: https://github.com/TUAN130294/awf/tree/main/awf_skills/awf-session-restore
Command: npx skills add https://github.com/TUAN130294/awf --skill awf-session-restore

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This skill optimizes AI session startups by lazily restoring context through three load levels.

Core Features & Use Cases

Three-level lazy-loading: Level 1 loads summary and current task; Level 2 loads decisions and pending tasks; Level 3 loads full history, errors, and checkpoints.
Efficient token budgeting: Minimizes token usage while preserving essential context for quick re-engagement.
Smart recap support: Enables quick resume of sessions with incremental detail on demand.

Quick Start

Run the awf-session-restore skill to lazily load your current session context. Use /recap to immediately load Level 1 data, or /recap full to load Level 1+2, or /recap deep for all levels.

Frequently Asked Questions about awf-session-restore

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

How do I restore AI session context without exceeding token budgets?▼

You can restore AI session context efficiently by using a three-level lazy-loading strategy that progressively loads summaries, decisions, and full history while enforcing a token-budget aware sequence to minimize token usage.

What is lazy-loading context restoration for AI workflows?▼

Lazy-loading context restoration is a technique that optimizes AI workflow startups by incrementally loading session data through multiple levels, allowing rapid re-engagement without immediately consuming the entire token budget.

How do I use recap commands to resume an AI session?▼

Use /recap to load Level 1 summary and current task data, /recap full to load Level 1 and 2 decisions, or /recap deep to load all levels including full history, errors, and checkpoints for immediate session resumption.

What's the best way to re-engage an interrupted AI work session quickly?▼

The best way to re-engage interrupted AI work quickly is to trigger a Level 1 context restoration, which loads only the session summary and current task to provide immediate, low-token-cost recapability.

Does lazy context restoration support deterministic loading behavior?▼

Yes, this approach enforces deterministic behavior and explicit level-based loading sequences, ensuring predictable performance and stable context restoration outcomes for complex AI workflows.