A
licenseNot graded
qualityB
maintenanceMCP server that intercepts and controls AI agent actions in your codebase by enforcing policies on file operations and commands, with logging, approval workflows, and rollback capabilities.
MIT
checkpointer
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@checkpointer undo my last edit to config.json"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
checkpointer
MCP server for file checkpointing and undo. Built with the Python MCP SDK v2 (mcp[cli] 2.0.0b1), implementing the 2025-07-28 MCP spec.
Lets an AI agent safely read, write, and edit files within a project directory while maintaining a full snapshot history. Changes can be reverted per-file or globally without requiring git or any external VCS.
How it works
All file state is tracked in a .checkpointer/ directory at the project root. Before any mutating tool runs, the current file content is hashed, compressed, and stored as a snapshot. A master ledger (master.json) records the per-file undo stack and a global edit history.
This adds minimal token overhead — snapshots are stored on disk, not in context.
Related MCP server: claude-vigil-mcp
Tools
Tool | Description |
| Read a file with line numbers. Smart-truncates large files (head+tail). Supports |
| List directory contents, optionally recursive to a given depth. |
| Overwrite or create a file. Snapshots before writing. |
| Replace a line range. Returns surrounding context with line numbers. |
| Content-based find-and-replace. |
| Revert a specific file to its state before the last mutation. |
| Revert whichever file was last mutated. |
| Discard all AI changes for a specific file. |
| Revert all AI changes across the project and delete history. |
| Unified diff of a file against its original state. |
Setup
Requires Python 3.10+.
uv syncSet PROJECT_DIR to the root of the project you want to checkpoint:
PROJECT_DIR=/path/to/project uv run main.pyConfiguration
After cloning the repo, add to your Claude Code MCP config:
{
"mcpServers": {
"checkpointer": {
"command": "uv",
"args": [
"run",
"--directory",
"/path/to/this/repo/",
"server.py"
],
"env": {
"PROJECT_DIR": "/path/to/target/project/"
}
}
}
}Available Tools
10 toolsedit_file_linesOverwrite a specific chunk in a file.ADestructive
To replace a specific chunk in a file from start_line inclusive to end_line inclusive with content(one based indexing for lines), after it's snapshot has been taken.
Inputs: content (required), end_line (required), filepath (required), start_line (required). Returns: result
get_session_diffB
Get a diff of what exactly was modified in a file from its orignal state.
Inputs: filepath (required). Returns: result
list_directoryGet the structure of a directory.ARead-only
To read the structure inside a directory, either surface level or recursively. Results will be truncated if they exceed 500.
Inputs: directory_path (required), depth, recursively. Returns: result
read_fileRead a file inside the project directory.ARead-only
To read the contents of a file with line numbers. Returns numbered lines. For large files, output is truncated to head+tail unless a specific range is given via start_line/end_line (1-based, inclusive).
Inputs: filepath (required), end_line, start_line. Returns: result
reset_sessionWipe all AI changes across the directory.DDestructive
Wipe all AI changes across the directory
Inputs: no named parameters. Returns: result
revert_fileUndo all LLM changes for this specific file.BDestructive
Discard all AI changes for this specific file.
Inputs: filepath (required). Returns: result
str_replaceReplace matching text in a file.ADestructive
Replace an exact occurrence of old_str with new_str in a file. old_str must appear exactly once in the file. No line numbers needed — matches on content.
Inputs: new_str (required), old_str (required), filepath (required). Returns: result
undo_file_editUndo changes in a file.ADestructive
Revert file before the most recent mutating tool was called on it. This will be recorded in the master ledger as an undo operation.
Inputs: filepath (required). Returns: result
undo_globalUndo changes in the last file edited or reverted.ADestructive
Revert the last file edited before the most recent mutating tool was called on it(undo_file_edit is also mutating)
Inputs: no named parameters. Returns: result
write_fileOverwrite a entire file.CDestructive
To completely overwrite a file, after it's snapshot has been taken. (or to create a new file)
Inputs: content (required), filepath (required). Returns: result
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
B3.1/5.0
Scored across 10 tools
Disambiguation4/5
Most tools have distinct purposes, but undo_file_edit and undo_global could be confused due to similar names and descriptions. edit_file_lines and str_replace both replace content, though their methods differ. Overall, the overlap is minor.
Naming Consistency4/5
Tool names follow a verb_noun snake_case pattern (e.g., edit_file_lines, list_directory, write_file). However, 'str_replace' deviates from the pattern (no underscore after verb), and 'undo_file_edit' is slightly awkward. Overall consistent.
Tool Count5/5
10 tools is appropriate for the scope of file editing and version control. Each tool has a clear role without being excessive or insufficient.
Completeness4/5
The tool set covers reading, writing, editing (two methods), listing, diff, undo, revert, and session reset. Missing features like manual snapshot saving or session listing are minor gaps given the automatic snapshotting implied.
Maintenance
ActivityInactive
ResponsivenessNo recent community issues
Related MCP Connectors
MCP server for building and testing AI agents with multi-model experimentation and insights.
Persistent file storage for AI agents via MCP and curl. Upload, download, and version files.
Persistent file storage for AI agents. Save files over MCP, get a link a human can open.
MCP server for AI agents to plan, verify, and deploy Cloudflare-native apps.
Related MCP Servers
- AlicenseAqualityDmaintenanceAn MCP server for creating named checkpoints and snapshots of projects, with selective file restore, content diffs, and artifact preservation, enabling safe recovery from external file changes in Claude Code.516 npm6MIT
- AlicenseNot gradedqualityDmaintenanceAn MCP server that provides on-demand safety for AI coding workflows, enabling inspection, review, checkpointing, and rollback of risky actions.37 npm1MIT
- AlicenseNot gradedqualityBmaintenanceAn MCP server that lets multiple AI coding agents share state, preserve context across sessions, and coordinate via a knowledge graph, messaging, and branch-preserving checkpoint/resume with conflict handling.57 npm1Apache 2.0