get_article_content
Get full article content and AI analysis for a bookmark by ID (same as get_bookmark)
add_to_collection
Add bookmarks to an existing Collection. Duplicates are silently ignored.
LABSCO SUMMARY
The deadline is the product, and the agent is what makes it survivable.
A 24-hour expiry only works if something triages the inbox, and that is precisely the chore nobody does. batch_triage_flame hands the decision to an assistant that has actually read the articles — and write_bookmark_analysis stores the reasoning it did instead of throwing it away when the session ends.
What it is
The MCP server for Burn, a read-later app with a deadline. Saves land in Flame with 24 hours to live, move to Spark for 30 days once read, and to Vault when they are worth keeping. These tools let an assistant triage, search, organise and analyse everything you have saved.
What you get
- Search and read — search_vault, list_vault by category, list_sparks, search_sparks, list_flame for what is about to burn, get_flame_detail, get_bookmark, get_article_content, list_categories
- Triage — move_flame_to_spark, move_flame_to_ash, move_spark_to_vault, move_spark_to_ash, and batch_triage_flame for a whole inbox in one call
- Collections — get_collections, get_collection_overview, create_collection, add_to_collection, remove_from_collection, and update_collection_overview for a written theme and synthesis
- Analysis — write_bookmark_analysis lets the agent analyse content with its own model and write structured results back, merging only the fields it supplied
- Fetching — fetch_content pulls an article or post from a URL, covering X, Reddit, YouTube, Bilibili, WeChat and ordinary web pages, checking the cache before going out
- Watched sources — add_watched_source for an X user, RSS feed or YouTube channel, list_watched_sources, remove_watched_source, and scrape_watched_sources to pull new items into Flame on demand or on a schedule
Requirements
A free Burn account and a token from Settings under MCP Server. Remote: point a client at https://burn-mcp-server.vercel.app/api/mcp with an Authorization: Bearer header — nothing runs locally. Local: the npm package burn-mcp-server over stdio with BURN_MCP_TOKEN in the environment; it exchanges that token for a short-lived session cached under your home directory, so it does not log in on every run. The free tier is 5 saves a day with all the MCP tools included.
Setup effort
One command plus a key — npx burn-mcp-server, then supply credentials
USE CASES
01
Clearing an inbox before it expires
See what is about to burn, deep-read the ones that matter, then decide the whole batch in one call.
02
Writing with your own sources
Find every relevant thing you kept, and pull the full article text to quote from.
03
Grouping saved reading into a theme
Create the collection, add the bookmarks, then write an overview of where the sources agree and disagree.
04
Following a person or a feed without a reader app
Watch an X account, RSS feed or YouTube channel; new posts arrive in the inbox on the next scrape.
05
Saving something you found outside the app
No tool adds a URL, so save it from the Burn app or Web Clipper first; then fetch the full content, including X, Reddit, YouTube or WeChat posts, and write the analysis back to that bookmark.
06
Finding what you read a month ago
Search across the 30-day window by keyword, tag or the takeaway recorded at the time.
26 tools
Counted once with only startup setup. Other versions, modes or optional credentials may expose a different list.
add_watched_source
Watch an X user, RSS feed, or YouTube channel — new posts auto-appear in Burn Flame on each scrape.
batch_triage_flame
Triage multiple Flame bookmarks at once. Each decision moves a bookmark to Spark or Ash.
create_collection
Create a new Collection to group related bookmarks together.
fetch_content
Fetch article/tweet content from a URL. Works with X.com (bypasses GFW via proxy), Reddit, YouTube, Bilibili, WeChat, and any web page. First checks Supabase cache, then fetches live.
get_bookmark
Get full details of a single bookmark including AI analysis and extracted content
get_collection_overview
Get a Collection by name with its AI overview and linked bookmarks metadata
get_collections
List all your Collections with article counts and AI overview themes
get_flame_detail
Get full details of a Flame bookmark including extracted article content, AI analysis, and reading guidance. Use this to deep-read a bookmark before deciding its fate.
list_categories
List all Vault categories with article counts
list_flame
List bookmarks in your Flame inbox (24h countdown). Shows AI triage info (strategy, relevance, novelty, hook) and time remaining. Use this to see what needs attention before it burns to Ash.
list_sparks
List your Sparks (bookmarks you have read, with 30-day lifespan). Includes spark insight and expiry date.
list_vault
List bookmarks in your Vault, optionally filtered by category
list_watched_sources
List all active watched sources (X users, RSS feeds, YouTube channels).
move_flame_to_ash
Burn a Flame bookmark to Ash (not worth keeping).
move_flame_to_spark
Move a Flame bookmark to Spark (mark as worth reading). Sets 30-day Spark lifespan.
move_spark_to_ash
Burn a Spark bookmark to Ash (not valuable enough to vault).
move_spark_to_vault
Promote a Spark bookmark to permanent Vault storage.
remove_from_collection
Remove bookmarks from a Collection.
remove_watched_source
Stop watching a source. Use list_watched_sources to find the source ID.
scrape_watched_sources
Fetch new content from all watched sources (or one specific source) and add new items to Burn Flame. Call this on a schedule or on demand.
search_sparks
Search your Sparks by keyword (searches title, tags, AI takeaway, spark insight)
search_vault
Search your Burn Vault for bookmarks by keyword (searches title, tags, AI takeaway)
update_collection_overview
Write an AI-generated overview for a Collection (theme, synthesis, patterns, gaps).
write_bookmark_analysis
Write AI analysis results into a bookmark. Agent analyzes content with its own LLM, then writes structured results back to Burn. Only provided fields are merged — existing data is preserved.
BEFORE YOU INSTALL
What to know first
Tools you get
26 tools when we connected. No tool saves new URLs or deletes bookmarks; the tools search, read, triage and organise what is already saved. The free tier includes all 26.
Requires
BURN_MCP_TOKEN in your environment
Where it runs
Of the 5 documented clients, 4 run the server on your machine and 1 connects to a hosted endpoint.
CONTEXT COST
Light Standing cost
Rated by the size of the 26 tool definitions that ride along on every turn, not by how many there are
INSTALLATION
Connect your client
Server setup
npx
TERMINAL
npx burn-mcp-serverHosted endpoint
ENDPOINT URL
https://burn-mcp-server.vercel.app/api/mcpAdd this hosted endpoint to your MCP client using the credentials listed under Requirements.
MCP Clients
Any MCP client — remote HTTP
ANY MCP CLIENT — REMOTE HTTP · JSON
{
"mcpServers": {
"burn": {
"type": "http",
"url": "https://burn-mcp-server.vercel.app/api/mcp",
"headers": {
"Authorization": "Bearer <YOUR_BURN_MCP_TOKEN>"
}
}
}
}Claude Desktop — npx, Cursor — npx, Windsurf — npx
JSON
{
"mcpServers": {
"burn": {
"command": "npx",
"args": [
"burn-mcp-server"
],
"env": {
"BURN_MCP_TOKEN": "<YOUR_BURN_MCP_TOKEN>"
}
}
}
}Claude Code
CLAUDE CODE · TERMINAL
claude mcp add burn -e BURN_MCP_TOKEN=<YOUR_BURN_MCP_TOKEN> -- npx burn-mcp-server
