How do I install the Checkpoint skill?

Run npx skillmds@latest add polyipseity/checkpoint in your terminal (requires Node.js), paste this page's agent-chat prompt into Claude, Cursor, or any MCP-connected agent, or download the SKILL.md file and copy it into your agent's skills directory.

Is Checkpoint safe to use?

This skill has not completed SkillMD's automated safety review yet. SkillMD never runs a skill's scripts for you; review the SKILL.md before installing.

Which AI agents work with Checkpoint?

This skill is tagged as working with Claude Code, Claude.ai, OpenAI Codex. SKILL.md is an open format, so most agents that read a skills directory can load it too.

Is Checkpoint free to use?

Yes. Installing skills from SkillMD is free, and the skill stays under its author's original license.

Who published Checkpoint?

polyipseity (@polyipseity) published this skill. Their other Agent Skills are listed on their SkillMD profile.

Checkpoint by polyipseity — Agent Skill

Name: Checkpoint
Author: polyipseity

File contents

Checkpoint: save session state

When to checkpoint

After completing each phase of an implementation plan
Before spawning subagents (saves state before delegation)
After receiving subagent results
When context is nearing capacity (long conversation, many tool calls)
Before any multi-edit batch

What to save

Write a checkpoint file to session memory at /memories/session/checkpoint-<datetime>.md.

Each checkpoint captures:

Step context: current plan step/phase, plan file path, frontmatter state
Work done: 2-3 sentence summary of what was accomplished this session
Next steps: what remains (list items from plan)
Cross-file state: file paths being edited, pending decisions, key search results

How to save

Memory tool availability: If the memory tool is not in the available tool list, call activate_vs_code_interaction with no arguments first — it is a one-shot call that permanently unlocks VS Code interaction tools.

Run date -u +%Y-%m-%dT%H%M%S for ISO timestamp.
Construct path /memories/session/checkpoint-<datetime>.md.
Call memory create with path /memories/session/checkpoint-<datetime>.md and file_text containing the checkpoint content.

How to restore

The continue.prompt.md, verify-plan.prompt.md, and verify-implementation.prompt.md workflows each find and load the latest checkpoint when resuming after interruption or context compaction. A checkpoint provides enough context to continue without replaying full history.

Integration

The plan.prompt.md and implement-plan.prompt.md prompts invoke this skill at natural break points. When those prompts invoke skill: "checkpoint", follow the save workflow above. Do not ask for confirmation.

---name: checkpointdescription: Save session state for context compaction survival. Called automatically during plan and implement-plan workflows at natural break points — after each phase, before subagent calls, and at context pressure.---# Checkpoint: save session state## When to checkpoint- After completing each phase of an implementation plan- Before spawning subagents (saves state before delegation)- After receiving subagent results- When context is nearing capacity (long conversation, many tool calls)- Before any multi-edit batch## What to saveWrite a checkpoint file to session memory at `/memories/session/checkpoint-<datetime>.md`.Each checkpoint captures:- **Step context:** current plan step/phase, plan file path, frontmatter state- **Work done:** 2-3 sentence summary of what was accomplished this session- **Next steps:** what remains (list items from plan)- **Cross-file state:** file paths being edited, pending decisions, key search results## How to save> **Memory tool availability:** If the `memory` tool is not in the available tool list, call `activate_vs_code_interaction` with no arguments first — it is a one-shot call that permanently unlocks VS Code interaction tools.1. Run `date -u +%Y-%m-%dT%H%M%S` for ISO timestamp.2. Construct path `/memories/session/checkpoint-<datetime>.md`.3. Call `memory create` with path `/memories/session/checkpoint-<datetime>.md` and `file_text` containing the checkpoint content.## How to restoreThe `continue.prompt.md`, `verify-plan.prompt.md`, and `verify-implementation.prompt.md` workflows each find and load the latest checkpoint when resuming after interruption or context compaction. A checkpoint provides enough context to continue without replaying full history.## IntegrationThe `plan.prompt.md` and `implement-plan.prompt.md` prompts invoke this skill at natural break points. When those prompts invoke `skill: "checkpoint"`, follow the save workflow above. Do not ask for confirmation.

Frequently asked questions

Checkpoint