Skip to content

Add MCP server endpoint for AI agent doc access - #2727

Open
joebutler2 wants to merge 3 commits into
freeCodeCamp:mainfrom
joebutler2:feature/mcp-server
Open

Add MCP server endpoint for AI agent doc access#2727
joebutler2 wants to merge 3 commits into
freeCodeCamp:mainfrom
joebutler2:feature/mcp-server

Conversation

@joebutler2

@joebutler2 joebutler2 commented Sep 7, 2026

Copy link
Copy Markdown

In keeping competitive with other API Doc services (like Dash), let's add MCP support so agents can interact with our Docs as well.

This adds POST /mcp implementing the Model Context Protocol (JSON-RPC 2.0, non-streaming HTTP) so AI coding agents can query a self-hosted DevDocs instance directly instead of scraping the browser UI.

Three tools, all reading from the same public/docs tree the server already uses to serve doc content:

  • devdocs_list_docsets — the configured doc sets (from settings.docs)
  • devdocs_search — entries in one doc set matching a query (index.json)
  • devdocs_get_page — one entry's content as plain text, HTML stripped with Nokogiri (already a dependency) (db.json)

Relates to #2420.

Adds POST /mcp implementing the Model Context Protocol (JSON-RPC 2.0,
non-streaming HTTP) so AI coding agents can query a self-hosted DevDocs
instance directly instead of scraping the browser UI.

Three tools, all reading from the same public/docs tree the server
already uses to serve doc content:
- devdocs_list_docsets — the configured doc sets (from settings.docs)
- devdocs_search — entries in one doc set matching a query
  (index.json)
- devdocs_get_page — one entry's content as plain text, HTML
  stripped with Nokogiri (already a dependency) (db.json)

Relates to freeCodeCamp#2420.
@joebutler2
joebutler2 marked this pull request as ready for review September 7, 2026 22:34
@joebutler2
joebutler2 requested a review from a team as a code owner September 7, 2026 22:34

@mo74m3ed mo74m3ed left a comment

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

- Add pagination support (offset/limit) to reduce response size
- Return condensed format (slug, name, version only) instead of full metadata
- Add query parameter for filtering by slug or name (case-insensitive)
- Include pagination metadata (offset, limit, total, returned) in responses
- Add comprehensive unit tests for all new features (7 new test cases)

All 11 MCP tests passing with 45 assertions.

Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>

Copilot AI left a comment

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

🟡 Changes recommended

Protocol interoperability, production storage, input security, and error-handling issues must be resolved.

Once you've addressed the issues Copilot identified, you can request another Copilot review.

Pull request overview

Adds an MCP endpoint so coding agents can query current DevDocs content, addressing #2420.

Changes:

  • Adds JSON-RPC initialization and tool discovery.
  • Adds docset listing, search, and page retrieval tools.
  • Adds MCP tests, fixtures, and local configuration exclusion.
File summaries
File Description
.gitignore Excludes local MCP configuration.
lib/app.rb Exposes the /mcp endpoint.
lib/mcp/server.rb Implements MCP request handling and tools.
test/mcp_test.rb Tests MCP operations.
test/files/docs/mcp_fixture/index.json Provides search fixtures.
test/files/docs/mcp_fixture/db.json Provides page-content fixtures.
Review details

Suppressed comments (2)

lib/mcp/server.rb:126

  • The hosted deployment provisions only meta.json locally (lib/tasks/docs.thor:263-266), not index.json, so every hosted devdocs_search call raises ENOENT. Use the configured documentation origin/storage backend or provision indexes during deployment.
      index_path = File.join(app_settings.docs_path, slug, 'index.json')
      index = JSON.parse(File.read(index_path))

lib/mcp/server.rb:126

  • The public request's slug can contain .., allowing this join to escape docs_path and inspect an unrelated index.json. Validate the slug against app_settings.docs and enforce containment within the expanded documentation root.
    def self.search_docset(app_settings, slug, query)
      index_path = File.join(app_settings.docs_path, slug, 'index.json')
      index = JSON.parse(File.read(index_path))
  • Files reviewed: 5/6 changed files
  • Comments generated: 9
  • Review effort level: Balanced

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

Comment thread lib/mcp/server.rb
case request['method']
when 'initialize'
respond(request, {
'protocolVersion' => '2024-11-05',
Comment thread lib/mcp/server.rb
Comment on lines +118 to +119
db_path = File.join(app_settings.docs_path, slug, 'db.json')
db = JSON.parse(File.read(db_path))
Comment thread lib/mcp/server.rb
Comment on lines +117 to +119
def self.get_page(app_settings, slug, path)
db_path = File.join(app_settings.docs_path, slug, 'db.json')
db = JSON.parse(File.read(db_path))
Comment thread lib/app.rb
Comment on lines +283 to +284
payload = JSON.parse(request.body.read)
Mcp::Server.handle(payload, settings).to_json
Comment thread lib/mcp/server.rb
Comment on lines +69 to +71
def self.call_tool(request, app_settings)
params = request['params']
case params['name']
Comment thread lib/mcp/server.rb
Comment on lines +71 to +81
case params['name']
when 'devdocs_list_docsets'
result = list_docsets(app_settings, params['arguments'] || {})
as_text_result(request, result)
when 'devdocs_search'
entries = search_docset(app_settings, params['arguments']['slug'], params['arguments']['query'])
as_text_result(request, entries)
when 'devdocs_get_page'
text = get_page(app_settings, params['arguments']['slug'], params['arguments']['path'])
respond(request, { 'content' => [{ 'type' => 'text', 'text' => text }] })
end
Comment thread lib/mcp/server.rb
Comment on lines +119 to +120
db = JSON.parse(File.read(db_path))
html = db[path]
Comment thread lib/mcp/server.rb
Comment on lines +120 to +121
html = db[path]
Nokogiri::HTML::DocumentFragment.parse(html).text.squeeze(' ').strip
Comment thread lib/mcp/server.rb
Comment on lines +127 to +130
query_lower = query.downcase
index['entries'].select do |entry|
entry['name'].downcase.include?(query_lower) || entry['path'].downcase.include?(query_lower)
end
- Validate docset slugs against configured docs to prevent path traversal attacks
- Handle missing index.json and db.json files gracefully in production
- Return descriptive JSON-RPC errors when files are unavailable
- Add tests for path traversal protection and missing file handling
- Add mcp_fixture to test docs manifest for proper test coverage

This addresses GitHub review concerns about:
1. Security: Path traversal vulnerability (slug with ..)
2. Production: Missing index.json and db.json in hosted deployments

Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants