Skip to content
Agentmart
Submit a tool

checkpoint

Human-in-the-loop review protocol.

Skill by chainreactors

IndexedMIT★ 26
Write a review

Install

Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.

npx agentmarket install skill/chainreactors-checkpoint

Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.

Get the weekly diff

The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.

We use your email only for this newsletter. Privacy

About

Human-in-the-loop review protocol. Use when execution reaches an approval boundary that requires human judgment before proceeding.

No reviews yet. Installed it? Reviews come from verified installers only. How reviews work

Permissions

What this skill can do and what it asks for, from its SKILL.md. Checked Oct 25, 2026.

Tools
No tools found
Secrets
None asked for
Install scripts
None found

Network hosts it talks to 1

  • codeload.github.com · found in install command

Files it touches

  • ~/.claude/skills/checkpoint

What's inside

Skill name
checkpoint
Description
Human-in-the-loop review protocol. Use when execution reaches an approval boundary that requires human judgment before proceeding.
License
MIT
Instructions
54 lines in SKILL.md · view on GitHub

Files (2)

  • SKILL.md
  • schema.json

SKILL.md (first lines)

# Checkpoint

Pause execution and submit an artifact for human review. Execution resumes only after the reviewer responds.

## Message format

Send with `content_type: "checkpoint"` on the message envelope.

Content body:

```json
{
  "id": "<unique-id>",
  "title": "Short description",
  "content": "Markdown body with full context for the reviewer",
  "options": ["Approve", "Reject"]
}
```

`id` is required and must be unique. Optional fields: `kind`, `target`, `status`.

Response — the reviewer sends this (with `refs.messages` pointing to your checkpoint):

```json
{
  "option": "Approve",
  "feedback": "Free-form comment from reviewer"
}
```

The response has **no `content_type`**. It is identified by the `refs.messages` relationship, not by content type.

## Mechanism

- `"Reject"` triggers mechanism-level rollback. All other options are semantic guidance.
- Submission blocks your session until feedback arrives.

## When to use

- Destructive or irreversible action ahead
- Plan needs approval before execution
- Human preference matters between alternatives

Do NOT checkpoint trivial decisions.

## After receiving feedback

Read both the option and the feedback text. The option is shorthand; the text carries nuance.

- Approve → proceed
- Reject → stop, roll back
- Custom option → interpret and adjust

From chainreactors/ioa, MIT.

Signals

Stars
26
Forks
4
Open issues
1
Last commit (this folder)
Jun 8, 2026 (4 months ago)
Last release
Jun 21, 2026 (4 months ago)

From GitHub (fetched Oct 24, 2026).

Details

License
MIT
First seen
Oct 24, 2026
Last checked
Oct 24, 2026

More from chainreactors

Cite this page

checkpoint skill. Agent Mart. https://agentmart.dev/skills/chainreactors-checkpoint (last updated 2026-10-24)