AI Signals (Signals API) — Claude Code MCP server: features and setup
This MCP server gives agents daily AI signals, ranking changes, and cited model and repository lists. Use it to check what changed since the prior snapshot.
claude mcp add --transport http signals-api-mcp https://data.jiangzhang.ca/signals4/mcpWhat AI Signals (Signals API) does
Five tools expose the daily signal set from GitHub, Hugging Face, YouTube, and Silicon Valley news; differences from the previous snapshot; ranked models and repositories with citations; and available archive dates. When a rank cannot be derived, the entity is marked unranked with a reason.
Connect over HTTP to the hosted MCP endpoint, or run the local mcp_server.py file. The local setup needs Python 3.9+ and an MCP client configuration. It uses no separate dependencies or API key. The free data feed has one daily edition and four sources. Direct API calls need a User-Agent header: the CDN responds with HTTP 403 to empty headers and Python-urllib.
The server code uses the MIT license. Data uses CC BY 4.0 and requires attribution to Signals 4 (Signals API). Ranking citations give agents a source for reported values instead of leaving them to invent one.
What it helps with
- retrieve the daily signals from four sources
- compare the latest snapshot with the previous one
- check model rankings and rank movement
- find the dates covered by the archive
Example prompts for the agent
- Show today's signals and link each one to its source.
- Compare the snapshots and report new entries and ranking changes.
- List the top models with citations for their ranks.
How to install AI Signals (Signals API)
Platform CLI: remote server
claude mcp add --transport http signals-api-mcp https://data.jiangzhang.ca/signals4/mcpPlatform CLI: remote server
codex mcp add ai-signals-signals-api --url https://data.jiangzhang.ca/signals4/mcpConfig file
{
"mcpServers": {
"ai-signals-signals-api": {
"url": "https://data.jiangzhang.ca/signals4/mcp"
}
}
}Config file
{
"servers": {
"ai-signals-signals-api": {
"type": "http",
"url": "https://data.jiangzhang.ca/signals4/mcp"
}
}
}Platform CLI: remote server
gemini mcp add -t http ai-signals-signals-api https://data.jiangzhang.ca/signals4/mcpStep by step, with commands for each platform: How to add an MCP server to Claude Code, Codex, Cursor, VS Code, Gemini CLI and more
Questions about AI Signals (Signals API)
How do I install AI Signals (Signals API) in Claude Code?
Run claude mcp add --transport http signals-api-mcp https://data.jiangzhang.ca/signals4/mcp. This connects Claude Code to the hosted server over HTTP.
What does AI Signals (Signals API) do?
It provides daily signals, changes between snapshots, and cited rankings for models and repositories. An archive tool lists available edition dates.
Does AI Signals require an API key or local Python?
The free service needs no API key. Python 3.9+ is required only for the local file; the hosted endpoint does not need local Python.
