local-comfyui-mcp

Overview Schema Source Related Servers Discussions Admin

restore_checkpoint

Restore recorded package versions to undo a bad upgrade, optionally removing newer packages for a full rollback. Requires stopping ComfyUI first and creates a safety checkpoint.

Instructions

Put a recorded set of package versions back. Takes a checkpoint of the current state first.

The undo for everything that installs. Two strengths, and the difference is what happens to packages that appeared since:

default - install the recorded versions and leave anything else alone. This is what undoes a bad upgrade, and it is safe because it only ever puts things back.
exact=True - also uninstall whatever is not on the recorded list. A true rollback, and the one that can lose work: everything it removes was installed after the checkpoint, which is usually a node pack's dependency and occasionally something that took an hour to compile. It is refused when the removal set holds a build no index can supply, and allow_removing is how a caller says it means it.

Not undoable, so a checkpoint of the current state is written first and named in the reply - the same guard load_workspace uses, and for the same reason: the tool cannot ask whether it is about to destroy something, so it makes sure the answer does not matter.

Refused while ComfyUI is running. Its interpreter holds the DLLs being replaced, and half-replaced packages under a live process is how an install stops starting. Call comfy_stop first.

Args: name: a checkpoint name from list_checkpoints, or an unambiguous prefix. exact: also remove packages that are not in the checkpoint. allow_removing: package names this call is permitted to remove even though nothing could install them back. Only meaningful with exact.

Input Schema

TableJSON Schema

Name	Required	Description	Default
`name`	Yes
`exact`	No
`allow_removing`	No

Output Schema

TableJSON Schema

Name	Required	Description	Default
No arguments

Schema Changelog

Changes observed during successful MCP inspections.

Addedv0.1.10Sep 13, 2026

TDQS

A4.9/5.0

Behavior5/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

With no annotations, the description carries the full burden and does so: it discloses the auto-written safety checkpoint, that exact mode can lose work and what it removes, that it is refused when a removal set holds an unsupplyable build, and the allow_removing escape hatch. This is far beyond what the name implies.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness4/5

Is the description appropriately sized, front-loaded, and free of redundancy?

Front-loaded with the core action, then bullets for the two modes, then the guards. Every sentence carries operational weight, though the prose is longer and more editorial than strictly necessary ('half-replaced packages under a live process is how an install stops starting').

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness5/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

Given an output schema exists, return values needn't be described, yet it notes the safety checkpoint is 'named in the reply.' Prerequisites, refusal conditions, destructive modes, and the recovery guard are all covered for a high-risk mutation tool.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters5/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema coverage is 0%, so the description must compensate, and it fully does: name is defined as a checkpoint name or unambiguous prefix from list_checkpoints, exact as 'also remove packages that are not in the checkpoint,' and allow_removing as package names permitted to be removed despite being uninstallable, noting it is only meaningful with exact.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

States a specific verb and resource ('Put a recorded set of package versions back') and immediately frames it as 'the undo for everything that installs,' which distinguishes it from create_checkpoint/list_checkpoints/delete_checkpoint siblings. An agent can tell what this does without opening the schema.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines5/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

Explicitly differentiates the two modes (default vs exact=True), states which scenario each fits ('undoes a bad upgrade' vs 'a true rollback'), and names the prerequisite alternative ('Call comfy_stop first') along with the refusal condition. When-to-use and when-refused are both spelled out.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

Deploy Server