survival-check

Assess PLAN.md task blocks for self-containment under context compaction.

Updated Jun 4, 2026
One-click install
npx skills add https://github.com/Kohlex/autopilot-plugin --skill survival-check
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: survival-check
Source: https://github.com/Kohlex/autopilot-plugin/tree/main/skills/survival-check
Command: npx skills add https://github.com/Kohlex/autopilot-plugin --skill survival-check

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

The Survival Check gate verifies that each PLAN.md task block is self-contained and can operate after a context compaction, preventing broken plans mid-execution.

Core Features & Use Cases

  • Self-contained evaluation: splits PLAN.md into per-task blocks and validates explicit file references and context pointers.
  • Plan-review gating: flags blocks that would rely on previous conversation or hidden state, surfacing blockers before execution.
  • Output artifact: emits a .survival-check.md with per-task verdicts and an overall PASS/FAIL to drive gating decisions.

Quick Start

Run the Survival Check on a PLAN.md to generate a .survival-check.md with per-task PASS/FAIL verdicts.

Frequently Asked Questions about survival-check

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

FAQPage Schema
How do I validate that plan tasks survive context compaction during autopilot execution?

To validate that plan tasks survive context compaction, run a survival check on your PLAN.md to split it into per-task blocks and verify each block is self-contained. It generates a .survival-check.md file with PASS/FAIL verdicts to gate execution.

What is a self-contained plan block in a plan-review workflow?

A self-contained plan block is a task segment that operates independently after context compaction by using explicit file references, context pointers, and contracts instead of relying on prior conversation history or hidden state.

How do I review PLAN.md files for hidden context dependencies before execution?

You can review PLAN.md files for hidden context dependencies by running a survival check gate that flags task blocks relying on previous conversation or implicit state, emitting a .survival-check.md artifact with an overall PASS or FAIL gating decision.

Can I use survival-check for plan-validation if my tasks reference prior outputs?

Yes, you can use survival-check for plan-validation when tasks reference prior outputs. It enforces explicit file references and context pointers within each PLAN.md task block to ensure they remain self-contained during context compaction.

Why does my plan break mid-execution after a context shift?

Plans break mid-execution after a context shift because task blocks lack explicit file references and context pointers, causing them to fail when prior conversation history or hidden state is compacted. A survival check gate prevents this by validating self-containment beforehand.

Do I need external dependencies to perform a plan-review survival check?

No external dependencies are required to perform a plan-review survival check. The tool independently splits PLAN.md into per-task blocks, validates their self-containment, and emits the .survival-check.md artifact with per-task verdicts.