validate_documentLABSCO SUMMARY
The paperwork of a repository, generated to a schema and checked against it.
Licenses and gitignores come from the same GitHub template repositories a human would copy from, so the output is the canonical text rather than something reconstructed. The document side is the more unusual half: the schemas are both the generator and the checker, which makes documentation completeness something an agent can audit instead of eyeball.
What it is
An MCP server for the boilerplate files every repository is expected to carry. It pulls license and gitignore templates from the official GitHub template repositories at call time, falling back to bundled copies when the API is unavailable, and it carries its own schemas for ten project documents — README, API, architecture, features, tech stack, TODO, backlog, changelog, contributing and prompts — which it can both generate and validate against.
What you get
- A license file written for a named license type, author and output path, drawn from 36 templates —
generate_license,list_license_templates - A
.gitignorefor a named stack, drawn from 155 GitHub templates —generate_gitignore,list_gitignore_templates - A starting document generated to the schema for its type, so a new README or architecture doc arrives with the expected sections already in place —
generate_template,list_schemas - An existing document checked against its schema, section by section —
validate_document - A whole project assessed for which documents exist and how complete they are —
analyze_project_docs - One call that lists everything available across schemas, licenses and gitignores —
list_templates - A fallback to local copies when the GitHub API cannot be reached, so generation still answers offline
Requirements
No account and no key — template fetches hit public GitHub endpoints. The README's run line names @scriptonbasestar/sb-schema-doc-server at 1.0.0, which npm does not have; the binary it documents is sb-schema-doc-server, spoken over stdio. Building from the clone needs Node.js 18.0.0 or newer. Generation tools take an output_path and write the file there, so point them at the repository you actually mean.
Setup effort
Build from source — git clone https://github.com/ScriptonBasestar/sb-mcp-server.git
USE CASES
01
Setting up a new repository's required files
Generate the license for the terms you have chosen and a .gitignore for the stack, both written straight to the paths you name, without leaving the editor to copy them from GitHub.
02
Finding out how thin your documentation actually is
Assess the project for which documents exist and how complete each one is against its schema, then generate a starting file for whichever is missing.
03
Keeping docs to one house structure across repositories
Validate a README or architecture document against the schema for its type and get back which sections are missing or out of order — the same check on every repository.
04
Picking the right template when you're not sure of its name
List the license or gitignore templates available and match on the name before generating, rather than guessing an identifier and getting a fallback copy.
9 tools
Listed in project docs. Your client may show a different list.
generate_templatelist_schemasanalyze_project_docsgenerate_licenselist_license_templatesgenerate_gitignorelist_gitignore_templateslist_templatesBEFORE YOU INSTALL
What to know first
Tools you get
9 tools named in the project's documentation.
Where it runs
The documented client runs the server on your machine.
INSTALLATION
Connect your client
Server setup
TERMINAL
git clone https://github.com/ScriptonBasestar/sb-mcp-server.git
cd sb-mcp-server/doc-server
npm install
npm run buildMCP Clients
Claude Desktop
CLAUDE DESKTOP · JSON
{
"mcpServers": {
"sb-schema-server": {
"command": "node",
"args": [
"<YOUR_PATH_TO_SB_MCP_SERVER>/doc-server/build/index.js"
]
}
}
}
