pronounce — Claude Code MCP server: what it does, how to install
An MCP server that gives AI agents the correct pronunciation of developer jargon, drawing on 1,903 entries with audio and cited sources.
claude mcp add pronounce -- uvx pronounce-mcpWhat pronounce does
The server answers "how do you pronounce X" with audio instead of a guessed phonetic spelling. The dataset holds 1,903 developer-jargon entries, 1,283 of them tied to a citable source such as a quote from the word's creator or an official document. Each entry carries a confidence tag: settled by the creator, agreed by the community, or still contested.
For words with two live readings, like GIF, SQL, GUI or JWT, the server reads both alternates in sequence so an agent can voice the disagreement rather than pick a side on its own. The same project also ships a separate Bash CLI and a Claude Code skill built on the same dictionary.
It starts with uvx pronounce-mcp, so a working uv installation is required.
What it helps with
- look up the correct pronunciation of an unfamiliar term
- have an agent voice a contested pronunciation instead of guessing one
- check how a word's creator actually says it
- get a citable source for a pronunciation used in documentation
- avoid mangling a project name on a call or in a recording
Example prompts for the agent
- How do you pronounce kubectl?
- Read out both pronunciations of JWT
- Find a source for how nginx is pronounced
How to install pronounce
Platform CLI
claude mcp add pronounce -- uvx pronounce-mcpuvx runs the command named after the package; if the README names another one, use uvx --from pronounce-mcp <command>
Platform CLI
codex mcp add pronounce -- uvx pronounce-mcpuvx runs the command named after the package; if the README names another one, use uvx --from pronounce-mcp <command>
Config file
{
"mcpServers": {
"pronounce": {
"command": "uvx",
"args": [
"pronounce-mcp"
]
}
}
}uvx runs the command named after the package; if the README names another one, use uvx --from pronounce-mcp <command>
Config file
{
"servers": {
"pronounce": {
"type": "stdio",
"command": "uvx",
"args": [
"pronounce-mcp"
]
}
}
}uvx runs the command named after the package; if the README names another one, use uvx --from pronounce-mcp <command>
Platform CLI
gemini mcp add pronounce uvx pronounce-mcpuvx runs the command named after the package; if the README names another one, use uvx --from pronounce-mcp <command>
Step 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 pronounce
How do I install pronounce in Claude Code?
Run claude mcp add pronounce -- uvx pronounce-mcp.
What does pronounce do?
It speaks the pronunciation of developer-jargon terms on request, drawing on a dictionary of 1,903 entries tagged by confidence and, where available, linked to a source.
How is this different from a web search?
Instead of forum arguments, it returns an IPA transcription and, for part of the dataset, a direct quote from the term's creator or an official document as the source.
