Saved Playbooks (Private Beta)

Teach repeated work once

Saved Playbooks lets you give a repeatable ToolYour workflow a name, safe defaults, and acceptance criteria. Your compatible AI host can then run that same recipe later instead of reconstructing the workflow from scratch.

Examples include a Website Launch Check, a Weekly Website Growth Report, or a Developer Ship Check.

What a Saved Playbook is

A Saved Playbook is a versioned reference to an existing ToolYour skill or workflow. It includes:

• A name, category, and optional description
• Safe input defaults, such as a site URL
• Acceptance criteria for the person or team
• A durable receipt for every run

It does not run arbitrary shell commands, make production changes, or take over Cursor, Claude Code, Codex, or another AI host. Your host still controls the workspace, Git, terminal, and approvals.

Availability

Saved Playbooks is currently a private beta for the default ToolYour MCP profile. It is not part of the public ChatGPT connector while that surface is being reviewed.

When enabled for your account, an agent can use:

1. create_saved_playbook to save an existing ToolYour skill or workflow.
2. run_saved_playbook to execute an active saved version with optional input overrides.
3. get_saved_playbook_run to retrieve the recorded result and gate.

Each definition change creates a new version. Earlier run receipts remain attached to the version that actually ran, so a later edit cannot rewrite past evidence.

Example

Save “Developer Ship Check” using the registered developer-ship-checklist workflow,
with https://example.com as the default URL. Activate it and run it before release.

The agent should return the playbook run receipt, its actual status, and the gate reported by ToolYour. A partial result is not a passing result.

See MCP Playbooks for the existing skill catalog and MCP Quickstart to connect ToolYour.