deepgram skills
COMMUNITYLABSCO SUMMARY
Six of them do real work: api is a full REST and WebSocket reference generated straight from Deepgram's own OpenAPI and AsyncAPI specs, covering both the Nova and Flux model families; docs routes a question to the right page of Deepgram's documentation; examples links out to full runnable integrations with third-party platforms — Twilio, LiveKit, LangChain, the Vercel AI SDK, Discord, and a dozen more named in the skill itself; recipes gives a single-feature, single-language snippet under 50 lines across Python, JavaScript, Go, .NET, Java, Rust, and the Deepgram CLI; starters clones a ready-to-run demo app; and setup-mcp configures Deepgram's own MCP server for whichever coding tool is asking. The seventh, template-skill, is the maintainers' own scaffold for adding a new skill later — its description literally reads "Replace with description of the skill and when to use it," and it ships in the same list as the working six.
This is for someone building against Deepgram's transcription, synthesis, or voice-agent APIs who wants their coding agent to stop guessing endpoint shapes. It is scoped to this one repository's "product contracts" — API reference, docs, starters, recipes, examples, MCP setup. Deepgram's language SDKs (Python, JavaScript, Java, Go, Rust, Swift, Kotlin, .NET) each publish their own separate set of skills — seven product skills plus a maintainer skill — in their own repository, and none of that SDK-specific material lives here.
READ THE FULL ANALYSIS
What each skill expects. Three of the seven need a Deepgram API key already in the environment (DEEPGRAM_API_KEY), one — setup-mcp — talks to Deepgram's CLI or its MCP server rather than the API directly, two run with nothing configured, and the last is the unfinished template, which our check could not classify as either.
No license is declared for this repository, so the usual open-source terms a reader might assume do not automatically apply here.
ALSO IN THIS PACKAGE
deepgram
Deepgram skills for AI coding tools: speech-to-text, text-to-speech, voice-agent, audio-intelligence, text-intelligence, and browser-agent on-ramps, plus API reference, documentation, starter apps, recipes, integration examples, the CLI, MCP server setup, and self-hosted deployment
deepgram-js-sdk
Deepgram JavaScript and TypeScript SDK skills for speech-to-text, text-to-speech, voice agents, and audio intelligence
deepgram-python-sdk
Deepgram Python SDK skills for speech-to-text, text-to-speech, voice agents, and audio intelligence
deepgram-java-sdk
Deepgram Java SDK skills for speech-to-text, text-to-speech, voice agents, and audio intelligence
deepgram-go-sdk
Deepgram Go SDK skills for speech-to-text, text-to-speech, voice agents, and audio intelligence
deepgram-rust-sdk
Deepgram Rust SDK skills for speech-to-text, text-to-speech, voice agents, and audio intelligence
deepgram-dotnet-sdk
Deepgram .NET SDK skills for speech-to-text, text-to-speech, voice agents, and audio intelligence
WHAT'S INSIDE
7 showing · 7 totalNothing else to set up — install it and go.
api
The exact details of every request you can make to Deepgram's speech services — the address to call, what to send it, what comes back — plus the seventeen mistakes that most often break an otherwise working setup.
docs
Say what you are trying to build with Deepgram — live captions, a speaking voice, a phone assistant, a copy running on your own servers — and this names the handful of documentation pages that actually cover it.
examples
Deepgram rarely stands on its own, so these are complete, runnable demos of it plugged into something else — a phone system, a chat app, a website framework, a voice-AI toolkit — with all the wiring already done.
recipes
One feature, one language, under fifty lines, ready to run: the shortest piece of working Deepgram code that does the single thing you asked about, and nothing else.
setup-mcp
Connects your coding editor — Claude Code, Cursor or Windsurf — to Deepgram: the full set of Deepgram commands if you have their command-line tool installed, and searchable documentation if you don't.
starters
A small demo app that already works — transcribing speech, reading text aloud, holding a conversation — in the language and framework you happen to use, to copy down and build your own thing on top of.
template-skill
HOW TO GET IT
npx skills add deepgram/skillsnpx skills add deepgram/skills --skill <name> --full-depthPick the skill name from the Skills tab — each entry there installs independently.
/plugin marketplace add deepgram/skills/plugin install deepgram@deepgram-agent-skills/plugin install deepgram-js-sdk@deepgram-agent-skills/plugin install deepgram-python-sdk@deepgram-agent-skillsTyped inside the agent's own prompt, not in a terminal. The marketplace lists 7 plugins; three are shown. Every one installs the same way, with its own name before @deepgram-agent-skills.