Skip to main content
Glama
faisalkcmvp

Spec Forge MCP

by faisalkcmvp
OverviewSchemaRelated ServersScoreDiscussions
TypeScript
Local

Spec Forge MCP

Minimal NestJS MCP server for extracting tables from documents into one JSON file.

Local Setup

Install dependencies and build the server:

npm install
npm run build

Create .env from .env.example and add your OpenRouter key:

AI_API_KEY=your-openrouter-api-key
AI_BASE_URL=https://openrouter.ai/api/v1
AI_MAX_CHUNK_CHARACTERS=40000
DEFAULT_TABLE_OUTPUT_PATH=output/document-tables.json
HTTP_HOST=127.0.0.1
HTTP_PORT=3000
HTTP_ALLOWED_ORIGIN=*

The document table tool currently uses this hardcoded OpenRouter model:

anthropic/claude-3-haiku

Related MCP server: Techtenstein PDF MCP

Run HTTP MCP

Start the local HTTP MCP server:

npm run start:http

The MCP endpoint will be:

http://127.0.0.1:3000/mcp

Health check:

curl http://127.0.0.1:3000/health

Connect Claude Desktop

Claude Desktop may not connect directly to local HTTP MCP URLs from claude_desktop_config.json. Use mcp-remote as the bridge: Claude Desktop talks stdio to mcp-remote, and mcp-remote talks HTTP to this server.

Keep this server running with:

npm run start:http

Then open Claude Desktop:

Settings -> Developer -> Edit Config

On macOS, the config file is usually:

~/Library/Application Support/Claude/claude_desktop_config.json

Add this server under mcpServers:

{
  "mcpServers": {
    "spec-forge-mcp": {
      "command": "npx",
      "args": [
        "-y",
        "mcp-remote",
        "http://127.0.0.1:3000/mcp",
        "--allow-http",
        "--transport",
        "http-first"
      ]
    }
  }
}

If your config already has other MCP servers, only add the spec-forge-mcp block inside the existing mcpServers object.

Fully quit and reopen Claude Desktop. After restart, Claude should show the extract_document_tables tool.

Direct HTTP Clients

Clients that support Streamable HTTP MCP can connect directly to:

http://127.0.0.1:3000/mcp

Example Prompt

Use extract_document_tables on /Users/faisal/Documents/example.pdf and save the JSON to /Users/faisal/Documents/tables.json

The tool will always generate one JSON file.

Docker

Build the image locally:

docker build -t spec-forge-mcp .

Run the container (loads variables from .env):

docker run --env-file .env -p 3000:3000 --rm spec-forge-mcp

Or use Docker Compose:

docker compose up --build

Make sure AI_API_KEY (and any other AI_* vars) are present in your .env before starting.

Available Tools

1 tool
extract_document_tablesExtract Document TablesA

Convert a PDF, DOCX, text, markdown, CSV, JSON, or HTML document into table data and write one JSON output file. The source can be a local file path or an HTTP/HTTPS URL.

ParametersJSON Schema
NameRequiredDescriptionDefault
documentIdNoOptional stable identifier for the document in the generated JSON.
outputPathNoOptional path for the single generated JSON file.
sourcePathYesPath to the source document on the local filesystem, or an HTTP/HTTPS URL (e.g. a raw GitHub link to a .md or .txt file).
maxChunkCharactersNoOptional maximum characters sent to the AI model per chunk.

Output Schema

ParametersJSON Schema
NameRequiredDescription
outputPathYes
sourcePathYes
tableCountYes

TDQS

A4.2/5.0
Behavior3/5

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

No annotations are provided, so the description must fully disclose behavior. It does state that the tool writes one JSON output file, making the side effect explicit. However, it omits details such as whether existing files are overwritten, whether directories are created, or what happens if no tables are found. The chunking behavior implied by the maxChunkCharacters parameter is also not explained, leaving some ambiguity in the tool's execution.

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

Conciseness5/5

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

The description is two sentences long, with no filler. It front-loads the primary action ('Convert...'), then adds source flexibility. Every word contributes to understanding the tool's purpose and capabilities, making it an exemplary concise description.

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?

The description covers the core purpose, input formats, and output behavior, and an output schema exists to explain return values. It lacks a note about chunking or AI model usage, which is relevant given the maxChunkCharacters parameter. However, for a tool with a clear scope and no siblings, it is largely complete and sufficient for an agent to select and invoke it.

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?

The schema already provides descriptions for all four parameters, giving a baseline of 3. The description adds value by explicitly listing the supported document formats (PDF, DOCX, text, markdown, CSV, JSON, HTML), which is more informative than the schema's example of .md and .txt. This enhances understanding of the sourcePath parameter beyond the schema text.

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?

The description clearly states the tool's function: converting a document into table data and writing a JSON output file. It enumerates supported input formats and source types, making the purpose specific and unambiguous. There are no sibling tools to conflict with, so the description successfully communicates what the tool does.

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?

The description provides clear context on when to use the tool: when a document (of supported formats) needs to be converted to tabular JSON. It mentions both local file paths and URLs, giving concrete usage scenarios. Since there are no sibling tools, explicit alternatives are unnecessary, but the description could have hinted at intended use cases or limitations (e.g., not for images).

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

Install Server
RayrunBuild, deploy, and operate MCP servers without becoming an infrastructure team.ray.runAd

TDQS

A3.9/5.0
Disambiguation5/5

With only one tool, there is no possibility of confusion between tools. The single tool has a clearly defined purpose and inputs, making its function unambiguous.

Naming Consistency5/5

The tool name 'extract_document_tables' follows a clear verb_noun pattern and is descriptive. With only one tool, there is no inconsistency to penalize.

Tool Count2/5

A single tool is too few for the apparent scope of a server named 'Spec Forge MCP'. The tool, while useful, is narrowly focused and does not justify a server-level presence on its own.

Completeness1/5

The server provides only one operation with no surrounding workflow support. There is no way to manage, list, or further process the extracted data, making the surface severely incomplete for any broader purpose.

Maintenance

ActivityStale
ResponsivenessNo issues

Resources

Unclaimed servers have limited discoverability.

Looking for Admin?

If you are the server author, to access and configure the admin panel.

Related MCP Connectors

Related MCP Servers

  • mcp-docparser

    Jasonzld
    A
    license
    Not graded
    quality
    D
    maintenance
    A lightweight document parser MCP server that enables Claude to parse PDFs, Word, Excel, images (OCR), and other document formats with support for chunking and metadata extraction.
    2
    MIT

  • Techtenstein PDF MCP

    sathvic-kollu
    A
    license
    Not graded
    quality
    C
    maintenance
    MCP server that gives your Claude, Cline, or Cursor session the ability to extract text, tables, and metadata from any PDF URL — including scanned PDFs via OCR.
    MIT

  • Xberg MCP Server

    jamon8888
    A
    license
    Not graded
    quality
    B
    maintenance
    MCP server that extracts clean text, tables, and structured data from documents, images, code, and audio files, supporting 97 formats with OCR, transcription, and code intelligence.
    MIT

  • Docling MCP

    saurabh-goyal-saurabh
    A
    license
    Not graded
    quality
    C
    maintenance
    A document processing MCP server that converts PDFs to structured formats and generates documents, with caching and RAG support.
    MIT

Latest Blog Posts

MCP directory API

We provide all the information about MCP servers via our MCP API.

curl -X GET 'https://glama.ai/api/mcp/v1/servers/faisalkcmvp/spec-forge-mcp'

If you have feedback or need assistance with the MCP directory API, please join our Discord server