Sombra MCP Server

Your AI agent can save, search, organize, and distill your web research through the Model Context Protocol.

Connecting

Recommended: set up via Claude.ai first

Adding Sombra as a connector in the Claude.ai or Claude Desktop app is the quickest way to get started. Once connected there, it automatically works in Claude Code too — no extra setup needed.

Claude Desktop & Claude.ai

Go to Settings → Connectors → Add Custom Connector
Enter "Sombra" as the name and set the URL to https://sombra.so/mcp
Click Connect, log in or create an account, and you're done

This connection carries over to Claude Code automatically.

Claude Code

If you set up via Claude.ai above, you're already connected. Otherwise, add the server directly:

claude mcp add --transport http sombra https://sombra.so/mcp

After adding, Claude Code will show Sombra as an unauthenticated connector. Select it and your browser will open to complete the OAuth login. Once you log in, the connection is active.

ChatGPT

In ChatGPT, go to Settings → Apps and enable Developer mode
Click Create App
Set the URL to https://sombra.so/mcp
Set auth type to OAuth — no other details needed
Save, log in when prompted, and you're done

Other MCP clients

Transport

Streamable HTTP

Endpoint

https://sombra.so/mcp

Auth

OAuth 2.1

Troubleshooting

"Not authenticated" in Claude Code

This means the OAuth flow hasn't completed yet. In Claude Code, run /mcp to see your connectors. Select the Sombra connector — your browser will open to complete login. Alternatively, add Sombra as a connector in the Claude.ai app first (Settings → Connectors), which also enables it for Claude Code.

Connection stopped working

MCP connections expire after 21 days of inactivity. Simply reconnect by selecting the Sombra connector again — no need to remove and re-add it.

Tools (19)

Save & Create

save_urlSave a web page by URL. Re-fetches and updates if previously saved.

create_artifactCreate a new note artifact with markdown content.

Pages that rely heavily on JavaScript or require authentication may not capture fully via URL alone. For best results with these pages, use the Chrome extension to capture them from your browser.

Read & Browse

browse_collectionsBrowse your collections returning title, id, and description.

read_collectionRead the full content from a specific collection.

fetch_artifactFetch a single artifact by ID.

browse_notesBrowse all note artifacts, sorted by most recent first.

search_artifactsSearch saved artifacts by title, content, or URL. Also searches collections.

list_archivedList all archived (deleted) artifacts, sorted by most recently archived.

read_collection_contextRead the distilled context/summary from a collection.

Update & Organize

update_artifactUpdate an artifact's metadata or content. Content editing for notes only.

move_artifactMove a single artifact to a different collection.

bulk_move_artifactsMove multiple artifacts to a different collection in one operation.

remove_from_collectionRemove an artifact from a collection (moves it back to unsorted).

create_collectionCreate a new collection to organize your saved content.

update_collectionUpdate a collection's name, description, or icon.

write_collection_contextWrite distilled context/summary to a collection. Preserves all code verbatim.

Delete & Restore

delete_artifactDelete an artifact (archives it, content is preserved).

delete_collectionDelete a collection (archives it, artifacts are preserved).

restore_artifactRestore a previously deleted artifact, making it active again.

Deletes are never permanent — all deleted artifacts and collections can be recovered with restore_artifact and list_archived.

Prompts (2)

Distillation

distill_collectionGenerate a context-engineered distillation prompt for a collection.

distill_technical_collectionSpecialized distillation for collections with heavy technical content.

Resources

Each collection is exposed as an MCP resource with the URI scheme sombra://project/{id}.

Start Free

Free forever · No credit card required