mcpField
F
failmodes

failmodes

Remote
github.com/elyngvedAI & Agents0Updated 16w ago
Repository

About failmodes

failmodes is an MCP server that provides a machine-readable taxonomy of AI system failure modes. Its tools let agents and other MCP clients search and retrieve structured records across categories such as factuality, alignment, planning, and code generation.

Key features

  1. 01Search by symptom or keyword
  2. 02Retrieve a full record by ID
  3. 03List taxonomy categories
  4. 04List modes with category filtering

Use cases

  1. 01Find failure modes matching an observed symptom
  2. 02Retrieve details for a known failure mode
  3. 03Browse modes within a taxonomy category

MCP support

Exposes four tools to MCP clients for searching, retrieving, and listing failure-mode taxonomy data. It is available over stdio or as a hosted Streamable HTTP endpoint.

Configuration

Remote servers need no local install. Pick your client, then copy the config or install in one click.

Endpointhttps://www.failmodes.com/api/mcp
Copy install command
Claude CodeTerminal
Cursor.cursor/mcp.json
VS Code.vscode/mcp.json
Claude Desktopclaude_desktop_config.json
Windsurf~/.codeium/windsurf/mcp_config.json

Run commands in a terminal; paste JSON into the client’s MCP config and restart. Servers that need sign-in will prompt on first connect.

Packages
npmfailmodes-mcp0.1.0

Tools

4
ParameterTypeRequiredDescription
symptomstringYesSymptom phrase or keyword to search for, e.g. 'made up a citation' or 'refused to help after pushback'

Details

From the project README

failmodes-mcp

MCP server for the Fail Modes taxonomy — a machine-readable knowledge base of AI system failure modes.

Provides four tools over 100+ canonical failure mode records spanning factuality, alignment, planning, code generation, instruction following, and more.

Tools

ToolDescription
search_failure_modesSearch by symptom text or keyword phrase. Returns ranked matches.
get_failure_modeRetrieve the full structured record for a failure mode by ID slug.
list_categoriesList all taxonomy categories with descriptions and mode counts.
list_modesList all modes, optionally filtered by category ID.

Quick Start

No installation required — run directly with npx:

npx failmodes-mcp

Configuration

Claude Desktop

Add to ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows):

{
  "mcpServers": {
    "failmodes": {
      "command": "npx",
      "args": ["-y", "failmodes-mcp"]
    }
  }
}
Cursor / VS Code

Add to your MCP settings file:

{
  "mcpServers": {
    "failmodes": {
      "command": "npx",
      "args": ["-y", "failmodes-mcp"]
    }
  }
}
Remote (no install required)

A hosted Streamable HTTP version is also available — no npm package needed:

{
  "mcpServers": {
    "failmodes": {
      "url": "https://www.failmodes.com/api/mcp"
    }
  }
}

How It Works

On first tool call, the server fetches current taxonomy data from failmodes.com and caches it in memory for the duration of the session. Requires internet access.

The stdio transport makes this compatible with any MCP client that supports running local processes (Claude Desktop, Cursor, VS Code, etc.).

Links

License

MIT

Comments

Markdown supported

New to MCP?

From what the protocol is to installing your first server, these answers cover most first questions.

An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.