Skip to main content
Glama
Local News Matters MCP server

Local News Matters MCP server

by rnakashima1

Local News Matters MCP server

An MCP server that exposes the articles and photos of Local News Matters — the Bay City News Foundation's nonprofit local-news site — to MCP-aware clients such as Claude Desktop and Claude Code.

It is a thin, read-only wrapper over the site's public WordPress REST API (/wp-json/wp/v2/), normalizing the verbose WordPress JSON into compact, LLM-friendly objects.

What it provides

Tools (search / read the live catalog)

Tool

Description

search_articles

Full-text article search with paging + filters (category, tag, author, date range).

list_recent_articles

The latest published articles.

get_article

A single article by numeric ID, including full body text.

get_article_by_slug

A single article by its URL slug.

search_photos

Search the photo/media library (images only).

get_photo

A single photo's metadata + every available size.

get_article_photos

All photos attached to a given article.

list_categories

Site sections, ordered by article count.

get_category

A single category by ID (name, slug, description, count).

list_tags

Topic tags (optionally filtered).

get_tag

A single tag by ID (name, slug, description, count).

list_authors

Contributors / authors.

list_pages

Static pages (About, Contact, Donate, …), sorted by title.

Every read tool above accepts an optional response_format argument — "json" (the default, raw structured data) or "markdown" (compact, human-readable text).

Resources (addressable by URI)

URI

Description

lnm://article/{id}

One article (full text) as JSON.

lnm://article/{id}/photos

Photos attached to an article.

lnm://photo/{id}

One photo's metadata and sizes.

lnm://recent/articles

The latest articles.

lnm://categories

The site's categories.

Because the archive holds thousands of items, the full catalog is reached through the search/list tools; resource templates address any individual item once you know its ID.

Related MCP server: citewire

Install & run

Requires Python 3.10+.

# clone the repo
git clone https://github.com/rnakashima1/localnewsmatters-mcp.git
cd localnewsmatters-mcp

# run it (uv installs dependencies on first run)
uv run localnewsmatters-mcp
# or, once installed into an environment
pip install .
localnewsmatters-mcp

The server speaks MCP over stdio, which is what desktop clients expect.

Configure a client

Claude Code

claude mcp add localnewsmatters -- uv run --directory /absolute/path/to/clone/localnewsmatters-mcp localnewsmatters-mcp

Claude Desktop (claude_desktop_config.json)

{
  "mcpServers": {
    "localnewsmatters": {
      "command": "uv",
      "args": ["run", "--directory", "/absolute/path/to/clone/localnewsmatters-mcp", "localnewsmatters-mcp"]
    }
  }
}

Configuration (environment variables)

Variable

Default

Purpose

LNM_BASE_URL

https://localnewsmatters.org

Site root; point at a staging mirror if needed.

LNM_USER_AGENT

localnewsmatters-mcp/0.1 …

Outbound User-Agent header.

LNM_TIMEOUT

30

Per-request timeout in seconds.

Development

uv sync          # install runtime + dev dependencies
uv run pytest    # run the test suite (fully offline; HTTP is mocked)

Notes & limitations

  • This is an unofficial client and is not affiliated with or endorsed by Local News Matters / Bay City News Foundation. Please respect the site's terms of use and use the data responsibly, with attribution.

  • The server depends on the WordPress REST API being publicly enabled at /wp-json/wp/v2/. If the site changes platforms or restricts the API, the endpoints in wp_client.py will need to be updated.

  • All access is read-only.

License

MIT

Available Tools

13 tools
get_articleA

Fetch a single article by numeric ID, including full body text.

Inputs: article_id (required), response_format. Returns: result

get_article_by_slugB

Fetch a single article by its URL slug (the last path segment of its URL).

Inputs: slug (required), response_format. Returns: result

get_article_photosB

List every photo attached to a given article.

Inputs: article_id (required), response_format. Returns: result

get_categoryA

Fetch a single category by ID, including its description and article count.

Inputs: category_id (required), response_format. Returns: result

get_photoB

Fetch a single photo by numeric ID, including caption, alt text and available sizes.

Inputs: photo_id (required), response_format. Returns: result

get_tagA

Fetch a single tag by ID, including its description and article count.

Inputs: tag_id (required), response_format. Returns: result

list_authorsC

List the site's authors/contributors.

Inputs: response_format. Returns: result

list_categoriesC

List the site's categories (sections), ordered by article count.

Inputs: response_format. Returns: result

list_pagesB

List the site's static pages (About, Contact, Donate, etc.), sorted by title.

Inputs: response_format. Returns: result

list_recent_articlesC

List the most recently published Local News Matters articles.

Inputs: per_page, response_format. Returns: result

list_tagsA

List topic tags, optionally filtered by a search term, ordered by usage.

Inputs: search, response_format. Returns: result

search_articlesA

Search Local News Matters articles.

Args: query: Full-text search terms. Leave empty to browse the most recent articles. page: 1-based page number for paging through results. per_page: Results per page (1-100). category_ids: Restrict to these category IDs (see list_categories). tag_ids: Restrict to these tag IDs (see list_tags). author_id: Restrict to a single author (see list_authors). after: ISO-8601 date; only articles published on/after it (e.g. "2024-01-01T00:00:00"). before: ISO-8601 date; only articles published on/before it. response_format: "json" (default) for raw data, or "markdown" for readable text.

Returns results (article summaries) plus total, total_pages, page.

Inputs: page, after, query, before, +5 more. Returns: result

search_photosA

Search photos/images in the Local News Matters media library.

Args: query: Search terms matched against title, caption and alt text. page: 1-based page number. per_page: Results per page (1-100). article_id: Restrict to photos attached to a specific article. response_format: "json" (default) for raw data, or "markdown" for readable text.

Returns results (photo summaries incl. source_url) plus total, total_pages.

Inputs: page, query, per_page, article_id, +1 more. Returns: result

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    TDQS

    B3.4/5.0

    Scored across 13 tools

    Disambiguation4/5

    Most tools have clearly distinct purposes, but there is some overlap: search_articles with an empty query duplicates list_recent_articles, and search_photos with article_id duplicates get_article_photos. The descriptions clarify these as convenience shortcuts, so an agent can still choose correctly, but the redundancy prevents a perfect score.

    Naming Consistency5/5

    All tool names follow a consistent verb_noun pattern in snake_case: list_authors, get_article, search_photos, etc. The few longer names like get_article_by_slug and get_article_photos still fit the same convention without deviation.

    Tool Count5/5

    With 13 tools covering articles, photos, authors, pages, categories, and tags, the set is well-scoped for a news site's read-only API. Each tool earns its place, and the count is comfortably within the ideal 3-15 range.

    Completeness4/5

    The surface covers the core read operations for all main resources: listing and fetching articles, photos, categories, tags, authors, and pages. Minor gaps include no get_author or get_page by ID, but these are unlikely to block typical agent workflows given the available list operations.

    Maintenance

    ActivityInactive
    ResponsivenessOutdated data

    Related MCP Connectors

    Related MCP Servers

    • A
      license
      Not graded
      quality
      C
      maintenance
      A read-only MCP server for searching ArcGIS Online content and querying hosted feature layers, allowing natural language interaction with AGOL items and feature services.
      MIT
    • A
      license
      Not graded
      quality
      F
      maintenance
      An attribution-first MCP server that provides tools to search and cite news and research metadata from free APIs and news platforms, without storing or republishing content.
      1
      MIT
    • A
      license
      Not graded
      quality
      C
      maintenance
      MCP server for WordPress content management via REST API, supporting posts, pages, media, comments, and terms through natural language interfaces like Cursor, ChatGPT, Codex, and Claude.
      MIT