executor-continuation

Save execution state to bookmarks and resume from exact checkpoints.

1|Updated Feb 22, 2026
One-click install
npx skills add https://github.com/gonz0w/bgsd-oc --skill executor-continuation
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: executor-continuation
Source: https://github.com/gonz0w/bgsd-oc/tree/main/skills/executor-continuation
Command: npx skills add https://github.com/gonz0w/bgsd-oc --skill executor-continuation

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Context window exhaustion can cause loss of work unless state is saved. This skill defines a robust protocol for saving execution state before exhaustion, resuming with a fresh agent, and verifying completion on resume.

Core Features & Use Cases

  • Saving and restoring state via a bookmark system that records phase, plan, and task indices, enabling seamless continuation.
  • Supports multi-phase plans and long-running tasks by allowing a new agent to pick up exactly where the previous one left.
  • Use cases include extended planning, execution with context constraints, and audit-friendly resumption.

Quick Start

Enable the continuation protocol by saving a bookmark before context exhaustion and resuming from the exact resume point in the next agent.

Frequently Asked Questions about executor-continuation

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

FAQPage Schema
How do I resume agent execution after hitting context window limits?

Agent checkpointing saves execution state via bookmarks that record phase, plan, and task indices before context exhaustion, enabling a fresh agent to resume the exact task without losing prior work.

When do I need to save execution state for long-running agent plans?

You need to save execution state for long-running agent plans when multi-phase tasks risk exhausting context windows, ensuring a new agent can seamlessly continue from a verified checkpoint on resume.

How does bookmark-based state saving work for agent continuation?

Bookmark-based state saving works by recording the current phase, plan, and task indices into a resume data structure, allowing a new agent to pick up exactly where the previous agent left off.

Can I use this continuation protocol for multi-phase task verification?

Yes, this continuation protocol supports multi-phase task verification by enabling a fresh agent to resume from a checkpoint and verify completion of the exact task before proceeding.

What happens to long-running tasks when context windows exhaust without state saving?

Without state saving, long-running tasks lose progress when context windows exhaust, but this continuation protocol prevents loss by saving bookmark-based resume data before the limit is reached.

Are there limitations to resuming an agent from a checkpoint?

A limitation is that the continuation protocol requires saving a bookmark before context exhaustion occurs; if the context window exhausts before state saving, the checkpoint data is lost and resumption fails.