A
licenseNot graded
qualityD
maintenanceA secure and scalable Git MCP server giving AI agents powerful version control for local and (soon) serverless environments.
91 npm
Apache 2.0
slivingdoc
by baalimago
This server provides a pull/commit workflow for syncing a shared notebook of plain text files between a local directory and a durable remote store, so multiple agents and humans can edit concurrently without overwriting each other.
notes_pull– writes the current notebook into a local directory (the server's default, or an absolutepathyou supply) and records the accepted state as the baseline for edits.notes_commit– publishes the changes made to the notebook directory since the last pull, and incorporates concurrent changes from others that don't conflict.Edit-between-sync model – slivingdoc only stores and merges; you (or an agent) do the actual writing of UTF-8 text files between a pull and a commit.
Merge-safe concurrency – independent edits from multiple clients land together; truly overlapping edits surface as a
CONTENT_CONFLICTcontaining both versions rather than silently clobbering one.Flexible directory targeting – omit
path(or pass an empty string) to use the server's own notebook directory, or point at any absolute host path up to 4,096 bytes.Commit messages – required on every commit; must be non-blank UTF-8 without U+0000 and at most 16,384 bytes, retained only in recent internal history.
MCP client integration – usable over stdio from any MCP-capable agent, with storage in your own S3-compatible bucket or a hosted space.
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., "@slivingdoccommit my notes with message 'updated project plan'"
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.
Test coverage: 88.9% 😍👌
Features
Durable context: automatically store agent's context
Merge-safe concurrent writes: move fast with no fear
Edits outsourced: slivingdoc concerns itself with storing, not writing
Simple interface:
pullandcommit, the rest is handled magicallyYour bucket or ours: any S3-compatible bucket, or a hosted space you can share with invite links
Powerful yet simple ACL: you decide where agents can read, where they can write
One binary: "batteries included", as they say
The docs cover
how it works,
the CLI and
configuration;
npx -y slivingdoc serve -h prints every flag.
Related MCP server: Cellar
Get started
Hosted. Sign in at slivingdoc.dev with GitHub or Google, create a token, and add the server to your agent:
claude mcp add slivingdoc \
--env SLIVINGDOC_TOKEN=<your-api-token> \
-- npx -y slivingdoc serveFree for one space, 10 MB and 250,000 requests a month (pricing).
Self-hosted. Point it at an existing S3-compatible bucket that supports conditional writes.:
export AWS_REGION=eu-north-1
export AWS_ACCESS_KEY_ID=<key>
export AWS_SECRET_ACCESS_KEY=<key>
npx -y slivingdoc serve --bucket my-notesNo bucket yet? examples/seaweedfs/ runs one
locally in a container, and terraform/ provisions one on AWS.
As stdio MCP client
See MCP hosts for how to configure the MCP server in your agentic coder. The same system applies in the openai agent SDK (and any other LLM engine with MCP client support).
As CLI
"Install" it via npx:
npx -y slivingdoc versionnpx fetches and verifies the native binary on first run. You can also
install the matching binary with the checksummed setup script:
curl -fsSL https://raw.githubusercontent.com/baalimago/slivingdoc/setup.sh | shIt installs into $HOME/.local/bin (or /usr/local/bin when run as root).
slivingdoc login
slivingdoc pull notes
echo "hello from $(hostname)" > notes/hello.md
slivingdoc commit notes -m "First note"Development
AGENTS.md is the developer guide and
architecture/ holds the contract, one concern
per file.
make qa # lint plus the full Go and npm test suitesLicense
MIT, see LICENSE. Third-party notices for the statically linked libgit2 are in NOTICE.
Not affiliated with Paris Hilton.
Available Tools
2 toolsnotes_commitA
Publish the caller's changes in the notebook directory and incorporate concurrent non-conflicting changes. Omit path or pass an empty string to use the server's notebook directory, which the result reports. message must be non-blank UTF-8 without U+0000, at most 16,384 bytes; it is retained in recent internal history only.
ParametersJSON Schema
| Name | Required | Description | Default |
|---|---|---|---|
| path | No | Optional absolute UTF-8 host path of the notebook directory, 1 through 4,096 bytes. Omit it or pass an empty string to use the server's notebook directory. | |
| message | Yes | Commit message: non-blank UTF-8 without U+0000, at most 16,384 bytes. |
TDQS
A3.6/5.0
Behavior3/5
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
With no annotations, the description carries the full burden. It does add real behavioral context beyond the name: default-path fallback, merge semantics for concurrent non-conflicting changes, and the notable fact that the message is retained only in recent internal history. However, it says nothing about permissions, failure/conflict behavior, or reversibility for what is effectively a mutation.
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?
Two sentences, front-loaded with the action and scope, followed by the constraints. The dense run of message constraints is necessary detail, though it makes the second sentence slightly overloaded.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Completeness4/5
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
For a two-parameter tool with no annotations and no output schema, the definition covers the default path, the merge behavior, and the message rules, and notes that the resolved directory is reported in the result. Error and conflict-resolution outcomes remain unstated, a minor gap.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Parameters4/5
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema coverage is 100%, so the baseline is 3, but the description genuinely adds meaning the schema lacks: the message's limited retention in internal history, and that the resolved notebook directory comes back in the result. These clarify semantics rather than restating constraints.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Purpose4/5
Does the description clearly state what the tool does and how it differs from similar tools?
States a specific verb+resource (publish changes in the notebook directory) and adds the merge scope (incorporate concurrent non-conflicting changes). It never names the sibling notes_pull, so an agent must infer the pairing from the name alone, which keeps it short of a 5.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Usage Guidelines3/5
Does the description explain when to use this tool, when not to, or what alternatives exist?
Usage is only implied: an agent can guess this is the write-side counterpart to notes_pull, but the description never says when to commit versus pull, nor any prerequisite such as needing local changes first. Minimum viable guidance, no explicit when/when-not.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
notes_pullA
Write the current notebook into the notebook directory and record the accepted state. Omit path or pass an empty string to use the server's notebook directory, which the result reports. Edit UTF-8 text files (without U+0000) there between notes_pull and notes_commit; notes_commit publishes the changes and incorporates concurrent non-conflicting changes.
ParametersJSON Schema
| Name | Required | Description | Default |
|---|---|---|---|
| path | No | Optional absolute UTF-8 host path of the notebook directory, 1 through 4,096 bytes. Omit it or pass an empty string to use the server's notebook directory. |
TDQS
A3.9/5.0
Behavior4/5
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
With no annotations, the description carries the full load and does disclose meaningful behavior: the path-default rule, the UTF-8/no-U+0000 constraint on editable files, the accepted-state recording, and that notes_commit merges concurrent non-conflicting changes. It omits whether existing local files are overwritten or what permissions are required, which a mutation-adjacent tool would ideally state.
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?
Three tight sentences with the primary action front-loaded and the default-behavior detail immediately after. Dense but every clause carries information; no filler.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Completeness4/5
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
For a one-optional-parameter tool with no annotations and no output schema, the description covers the default path, the edit window, file-format constraints, and the sibling's role. Return values are only partially addressed (it notes the result reports the directory), which keeps it from a 5.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Parameters3/5
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema coverage is 100% and the single path parameter is fully documented there, including the 1–4,096 byte range. The description only restates the omit/empty-string default and adds that the result reports the directory, so baseline 3 is correct.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Purpose4/5
Does the description clearly state what the tool does and how it differs from similar tools?
States a specific action (materialize the current notebook into a notebook directory and record the accepted state) and clearly distinguishes itself from the sibling notes_commit, which 'publishes the changes.' The phrasing 'write the current notebook into the notebook directory' is slightly ambiguous about data direction, but the pull/commit pairing makes the intent recoverable.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Usage Guidelines4/5
Does the description explain when to use this tool, when not to, or what alternatives exist?
Gives an explicit workflow: edit the UTF-8 files 'between notes_pull and notes_commit,' with notes_commit handling publication and merge. It does not state when not to use it (e.g. if local edits already exist) or what happens on re-pull, so it stops short of full when/when-not guidance.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
2 tool updates
v0.1.0- First observed
notes_commit - First observed
notes_pull
TDQS
A3.7/5.0
Scored across 2 tools
Disambiguation5/5
notes_commit and notes_pull have clearly distinct purposes: one publishes local changes and incorporates concurrent changes, the other writes the current notebook into the directory. Their descriptions make the boundary unambiguous.
Naming Consistency5/5
Both tools follow a consistent notes_<verb> snake_case pattern. There are no deviations or mixed conventions.
Tool Count3/5
With only two tools, the set is borderline thin for a notebook version control system. While each tool earns its place, typical workflows require more operations.
Completeness2/5
The toolset covers only pull and commit, missing critical operations such as conflict resolution, status, history, and initialization. These significant gaps will cause agent failures in realistic use.
Maintenance
ActivityActive
ResponsivenessNo issues
Related MCP Connectors
A MCP server built for developers enabling Git based project management with project and personal…
System-of-record notebook for AI coding agents: pages, datastores, tasks, skills over MCP.
An MCP memory server. One memory your agents share — across models, devices and apps.
Control plane for autonomous software labor. Agents claim objectives over MCP with audit trail.
Related MCP Servers
- AlicenseNot gradedqualityAmaintenanceAn interactive notebook server that enables humans and AI agents to collaboratively edit and run Jupyter notebooks on a shared kernel, with real-time streaming and automatic MCP integration for agent access.18 npmMIT
- AlicenseNot gradedqualityCmaintenanceA storage-backed MCP server for sharing durable project context between trusted coding agents, with project discovery, optimistic concurrency, and timestamped handoffs.Apache 2.0
- AlicenseNot gradedqualityAmaintenanceAn MCP server that gives AI agents a full, safe interface to Git with automatic backups, isolated worktrees, and structured JSON outputs.43MIT