Click on "Install Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type @ followed by the MCP server name and your instructions, e.g., "@lightning-fm-mcp What's currently playing on Lightning FM?"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
An MCP server for Lightning FM's music catalog: search tracks, look up artists, and check what's live on the station, all from an agent.
There is no separate agent API. This server reads exactly what a human client reads: kind 31337 track events and kind 0 profiles from public Nostr relays, audio and images from the public Blossom server, and the kind 30311 live activity event the station publishes on every track change. Every track is signed by the artist's own key. This is the canonical catalog, not a curated subset the platform maintains for agents. Artists keep their exit rights: the catalog lives in events they signed, and nothing this server returns depends on the platform's goodwill.
Agent developers: you want the tools in your MCP client. Go to Use it and Tools. No install, no API key, no account.
Contributors: you want to change or extend the server. Go to What works today and Development. The wire format this server parses is documented at lightning.fm/interop.
Related MCP server: Maximum Sats MCP
Add it to your MCP client's server config. For Claude Code or Claude Desktop:
{
"mcpServers": {
"lightning-fm": {
"command": "npx",
"args": ["-y", "lightning-fm-mcp"]
}
}
}No install, no API key, no account. It talks to relay.lightning.fm and media.lightning.fm by default. See Configuration to point it elsewhere.
search_catalog: free-text search plus genre/tag filters over the full catalog. Leave every filter empty to browse newest-first.
get_artist: look up an artist by display name (partial match), hex pubkey, or npub. Returns their profile and full discography.
get_now_playing: what's currently on air, and the stream URL.
Verified against src/, not against ambition.
Feature | Status |
| Works |
| Works |
| Works |
Read-only guarantee: the server holds no keys, never signs, never publishes; there are no write tools | Works, by construction |
Relay snapshot caching (default 60s TTL) so repeated tool calls don't hammer relays | Works |
Transport | stdio only; no HTTP transport yet |
Writes of any kind (publishing tracks, zapping, tipping) | Not built, and out of scope by design |
All environment variables are optional; defaults point at production. Nothing here is a secret. This server is read-only and never signs or publishes anything.
Variable | Default | What it does |
|
| Comma-separated relay URLs to query. Only add a relay here if it gates writes the way relay.lightning.fm does. Kind 31337 isn't exclusive to Lightning FM, and an unrestricted public relay hands back every other app's tracks too. |
|
| Blossom server audio/image hashes resolve against. |
|
| How long a fetched snapshot is reused before the next tool call re-queries the relays. |
Nothing about you, ever. This server has no telemetry, no analytics, no phone-home, and no opt-in flag for any of the above. It reports nothing to Lightning FM.
The one thing it does send is a User-Agent header on its outbound relay
connections:
User-Agent: lightning-fm-mcp/<version>That is the software naming itself. It is byte-identical for every install,
contains no identifier, and says nothing about you or your machine. Without
it the underlying websocket library sends its own default User-Agent
instead, so this replaces one generic string with a specific one; it does
not add a channel. It exists so Lightning FM can tell agent catalog queries
apart from browser traffic in its own relay logs.
If you would rather not send it, npm and a one-line edit to src/config.ts
are all it takes. It is a measurement aid, not a control.
npm install
npm run dev # runs src/index.ts directly via tsx
npm run build # tsc, emits dist/
npm test # tsx --test tests/*.test.tssrc/index.ts server entry, stdio transport
src/tools.ts the three MCP tool registrations
src/nostr/ relay reads: catalog.ts, profiles.ts, now-playing.ts, and track.ts (kind 31337 parsing)
src/config.ts env handling and defaults
tests/ parsing and catalog tests
src/nostr/track.ts mirrors the kind 31337 parsing in station-server/src/catalog/track.ts (private repo) and app-desktop/src-tauri/src/relay.rs. All three must agree on tag names. This is the platform's public wire format, documented in full at lightning.fm/interop.
MIT
Look up an artist by display name or Nostr pubkey (hex or npub) and return their profile plus every track they have published to Lightning FM.
| Name | Required | Description | Default |
|---|---|---|---|
| artist | Yes | Artist display name (partial match ok), hex pubkey, or npub. |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
With no annotations provided, the description takes on the full burden. It discloses accepted input formats (display name, hex, npub) and return contents (profile + every published track). It does not describe error handling or what happens on ambiguous matches, and the read-only nature is implied rather than explicit, so a 5 is not warranted.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
A single, efficiently worded sentence that front-loads the action and target, then conveys return value without filler. Every phrase earns its place.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
For a single-parameter lookup with no output schema, the description adequately covers what will be returned (profile + tracks). It leaves a few edge cases unaddressed (e.g., ambiguous names, empty results), but the tool is simple enough that these are minor gaps.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema description coverage is 100% and the schema already explains the parameter well (partial match ok, formats). The description essentially repeats that information without adding extra semantic context about the parameter, so it meets the baseline but does not exceed it.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description clearly states a specific verb ('Look up'), the resource ('artist'), and the accepted identifiers (display name or Nostr pubkey), and further specifies that it returns the profile plus all published tracks. However, it does not explicitly differentiate from sibling tools like search_catalog, so it stops short of a 5.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The intended use is clear: when you need an artist's profile and tracks by name or pubkey, this is the tool. It does not explicitly address when not to use it or mention alternatives, leaving sibling differentiation to inference.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Read the current live activity (NIP-53, kind 30311) Lightning FM publishes on every track change: the track currently on air, the stream URL, and station status.
| Name | Required | Description | Default |
|---|---|---|---|
No parameters | |||
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
With no annotations, the description carries the full burden. It clearly labels the operation as a read, exposes the live/NIP-53 protocol detail, and states what information is returned. It does not mention failure modes, latency, or auth, but for a zero-parameter public read that is a minor gap.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is a single tight sentence that front-loads the verb and resource, then adds protocol and return details. Every clause earns its place with no repetition or fluff.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
For a zero-parameter read with no output schema, the description needs to explain what the tool returns and does, and it does so concretely (track, stream URL, station status). It could slightly improve by noting what happens when no track is currently on air, but the station status detail partially covers that.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
The tool takes zero parameters, so the schema holds no parameter information for the description to enhance. According to the baseline for zero-parameter tools, this is a Strong case: nothing is missing.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description names a specific verb ('Read'), a specific resource ('current live activity ... Lightning FM publishes'), and the exact data elements returned (track, stream URL, station status). It clearly distinguishes this from sibling search-oriented tools like search_catalog and get_artist by focusing on live, current state rather than catalog lookup.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The description says it reads current live activity and the track currently on air, which gives clear context for when to use it. It does not explicitly name sibling alternatives or say when not to use it, but the 'current' vs. search distinction is strong enough that an agent should route correctly.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Search or browse Lightning FM's Nostr-native music catalog (kind 31337 track events, fetched live from wss://relay.lightning.fm). Every result is signed by the artist's own Nostr key — this is the canonical catalog, not a curated subset. Leave all filters empty to browse the whole catalog, newest first.
| Name | Required | Description | Default |
|---|---|---|---|
| tag | No | Match a single free-form keyword tag. | |
| genre | No | Exact or partial genre match, case-insensitive. | |
| limit | No | Max tracks to return (default 20, max 100). | |
| query | No | Free-text match against title, artist name, album, genre, and description. |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
With no annotations provided, the description carries the full burden of behavioral disclosure. It discloses the live-fetching behavior, the Nostr signature guarantee, the canonical nature of the data, and the default ordering. It does not mention potential latency, relay availability, or result shape, but for a search tool the disclosed traits are substantive and useful.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is three sentences with no filler. It front-loads the core purpose, then efficiently adds key contextual facts (live source, signed results, canonical scope, browse behavior). Every sentence earns its place.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
For a search/browse tool with no output schema and no annotations, the description covers purpose, usage, data source, trust properties, and default ordering. It is missing explicit return-format details and error behavior, but the schema covers parameters and the description gives enough context for correct invocation.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema description coverage is 100%, so all parameters are already documented in the schema. The description adds one meaningful behavioral hint—that leaving all filters empty browses the whole catalog—but does not add meaning for individual parameters beyond the schema. Baseline 3 is appropriate.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description states a specific verb ('Search or browse') and resource ('Lightning FM's Nostr-native music catalog'), and further distinguishes it by specifying the event kind (31337) and the live relay source. This makes it clearly distinct from the sibling tools get_artist and get_now_playing, which focus on singular or current-state data.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
It gives clear usage context: filters can be left empty to browse the entire catalog newest first, and the catalog is live and canonical rather than curated. However, it does not explicitly state when to prefer this tool over the sibling tools or provide exclusion criteria, so it stops short of a 5.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
v0.1.5get_artistget_now_playingsearch_catalogScored across 3 tools
Each tool has a distinct purpose: searching the catalog, looking up an artist, and reading live now-playing data. There is no practical overlap in what an agent would select them for.
All three tools follow a clear verb_noun convention: search_catalog, get_artist, get_now_playing. The naming pattern is predictable and consistent.
Three tools is a tight, well-scoped set for a read-only music catalog API. Each tool covers a distinct high-level user need without unnecessary bloat.
The core domain is covered: catalog browsing/search, artist lookup with tracks, and live playback status. A direct track-by-ID detail lookup is a minor gap, but the other tools can work around it through search or artist queries.
Unclaimed servers have limited discoverability.
If you are the server author, to access and configure the admin panel.
The media memory layer for AI agents and their humans. Your AI client gets 29 tools to search your collection, add items, update ratings, preview music, and find patterns across everything you've read, watched, and listened to.
Find Bizzy radio stations, read station cards and public author information. Read-only; no playback.
Agent registry with Nostr identity, reputation, escrow, observability, and Lightning payments.
Search ATProto writing, annotations, identity, agents, and forum posts. 12 read-only tools.