changologs-mcp

changologs-mcp

An MCP server for Changologs. It lets Claude and other MCP clients create, search and edit your logs, and manage your tasks.

It's a hosted, remote server. You don't install anything: connect your client to the URL below and sign in with your Changologs account.

https://mcp.changologs.com/mcp

Listed in the MCP Registry as io.github.altyaper/changologs-mcp.

Connect

Claude Code

claude mcp add --transport http changologs https://mcp.changologs.com/mcp

Then run /mcp inside Claude Code and choose changologs to sign in.

Claude Desktop / claude.ai

Open Settings → Connectors → Add custom connector, and enter https://mcp.changologs.com/mcp as the URL.

Other clients

Any client that supports streamable HTTP transport and OAuth can connect to the same URL.

Authentication

The server uses OAuth 2.0 with PKCE, and Changologs is the authorization server. The client finds the auth server's endpoints at /.well-known/oauth-authorization-server and registers itself automatically (RFC 7591 dynamic client registration). It then opens a Changologs sign-in page in your browser. After you approve, every request carries your access token, so the server can see only your own workspaces, logs and tasks.

The server doesn't store anything. Each request goes straight to the Changologs API with your token.

Tools

Logs

Tool	What it does
create_log	Create a new log (new_log: true, or whenever a title is given), or append to today's daily note (the default for untitled quick notes).
list_workspaces	List your workspaces and their ids, for create_log.
list_recent_logs	List the most recently updated logs.
search_logs	Search logs by title or content.
get_log	Get a log's full content, by title.
update_log	Append to or replace a log's content, by title.
set_log_icon	Set or remove the emoji shown next to a log's title.
delete_log	Delete a log, by title.

Rich content blocks

These render an HTML fragment matching the changologs editor's formatting exactly — they don't call the API. Concatenate the fragments in order and pass the result as text to create_log/update_log to produce a richly formatted log.

Tool	What it does
add_paragraph	A paragraph. Supports **bold**, _italic_, `code`, [text](url).
add_heading	A heading, level 1-3.
add_bullet_list / add_numbered_list	A bullet or numbered list.
add_task_list	A checkbox list (inline content, not a create_task_list task list).
add_blockquote	A blockquote.
add_code_block	A fenced code block, optionally with a language label.
add_divider	A horizontal rule.
add_toggle	A collapsible section with a summary and hidden content.
add_callout	A highlighted callout box with an icon.
add_table	A table. No header-row styling exists in this editor, so header_row: true bolds the first row's text instead.
add_image	An image, from an already-hosted URL (no file upload).

Tasks

Tool	What it does
list_task_lists	List your task lists and their ids, for create_task.
create_task_list	Create a task list, optionally in a given workspace.
list_tasks	List the tasks in a list, in display order. Completed tasks are optional.
create_task	Create a task, with optional details, date and deadline.
get_task	Get a task's details, by title.
update_task	Rename a task, or change its details, date or deadline.
complete_task	Check off a task, or uncheck it. Checking a repeating task moves it to its next occurrence.

Databases

A database is a log whose sub-logs are rows in a table: each row's title is its Name, and the other columns are typed (text, number, date, select, multi-select, checkbox, url, email, phone, person). Databases are looked up by title within a workspace; rows by title or by the row_id that query_database_rows shows.

A row's emoji is the row log's icon, which changologs draws next to the title; it is set with the icon parameter, never by writing the emoji into the title. An emoji at the start of a title is moved to the icon, so a row does not end up showing it twice.

Tool	What it does
list_databases	List the databases in a workspace.
get_database	Show a database's columns (with select choices) and views (with filters and sorts).
query_database_rows	List rows as a table, filtered and sorted by a view, 100 per page.
create_database	Create a database, optionally with its columns.
convert_log_to_database	Turn a log into a database. Its sub-logs become rows, with columns built from their properties.
add_database_column / update_database_column / delete_database_column	Add, rename, retype, re-choice or delete a column.
add_database_row	Add a row with a title, an optional emoji icon and cell values keyed by column name.
update_database_row	Change a row's title, its icon or some of its cells.
delete_database_row	Move a row to the trash.
update_database_view	Replace a view's filters and sorts. This changes the saved view in changologs for everyone.

Example prompts

• "Log what we did in this session to Changologs."
• "Write a new log with all this data in a table, add callouts where it makes sense, and use h1/h2/h3 headings and paragraphs."
• "Search my logs for the Postgres migration notes."
• "Add a task to finish the release notes by Friday."
• "What's left on my Work task list?"
• "Add a row to my Reading list database for Dune, status Reading."
• "Show me every row in the Bugs database that isn't Done."

Development

The hosted server is a Cloudflare Worker (worker.ts). server.ts is a standalone Node version of the same server, for local development.

npm install

# Worker, locally
npm run dev:worker

# Node server on http://localhost:3001/mcp, pointing at a local Changologs app
CHANGOLOGS_BASE_URL=http://localhost:3000 npm start

# Type-check
npm run build

# Run tests
npm test

The Worker reads CHANGOLOGS_BASE_URL from wrangler.json (set to https://changologs.com). npm run deploy deploys it with Wrangler.

To update the registry listing, bump version in server.json and run mcp-publisher publish.