examples

Agent-maintained examples showing how to use Deepgram SDKs with popular platforms, frameworks, and ecosystems.

All examples are built and maintained by autonomous agents. Humans can direct, override, and add examples at any time.

→ Contributing · → Open PRs · → Request an example

How it works

1. Discover — Agents scan for new platforms and ecosystems weekly, queuing example ideas
2. Create — Agents build working example apps with tests and documentation, raising PRs
3. Review — Agents self-review their PRs; Copilot review also runs automatically
4. Test — Language-specific CI workflows run the example's tests
5. Fix — If tests fail for fixable reasons, the fix agent pushes a repair and tests retry
6. Merge — Auto-merge activates once all checks pass

If an example requires external credentials that aren't configured as repository secrets, the PR stays open with a comment tagging @deepgram-devrel listing exactly what's needed. Every example in the commit history represents a complete contribution → review → fix → merge lifecycle.

Examples

#	Example	Language	Integration	Status
010	Getting Started — Transcribe a URL with Node.js	Node.js	Deepgram SDK	✅ passing
020	Twilio Media Streams — Real-Time Call Transcription	Node.js	Twilio Media Streams	❌ failing
030	LiveKit Agents — Voice Assistant with Deepgram STT	Python	LiveKit Agents	❌ failing
040	LangChain STT Tool — Transcribe Audio in AI Pipelines	Python	LangChain	❌ failing
050	Vercel AI SDK — Transcribe Audio and Generate Speech with Deepgram	Node.js	Vercel AI SDK	✅ passing
060	Discord Bot — Transcribe Audio Attachments with Deepgram	Node.js	Discord	✅ passing
070	Vonage Voice API — Real-Time Call Transcription	Node.js	Vonage Voice API	✅ passing
080	Pipecat Voice Pipeline — Conversational Bot with Deepgram STT & TTS	Python	Pipecat	❌ failing
090	React Native Live Transcription	Node.js	React Native	✅ passing
100	FastAPI Audio Transcription API	Python	FastAPI	❌ failing
110	Cloudflare Worker — Edge Audio Transcription	Node.js	Cloudflare Workers	✅ passing
120	Slack Bot — Auto-Transcribe Audio Messages with Deepgram	Node.js	Slack	⏳ needs credentials
130	Telegram Voice Transcription Bot	Python	Telegram	⏳ needs credentials
140	Audio to Subtitles CLI	Python	Deepgram SDK	❌ failing
140	Flutter Voice Transcription	Dart	Flutter	—

Directory structure

examples/
  {NNN}-{slug}/           # e.g. 010-getting-started-node, 020-twilio-voice-agent-node
    README.md             # What it does, prerequisites, env vars, how to run
    .env.example          # Every required environment variable (no values)
    src/                  # Source code
    tests/                # Tests — must exit 0 on success, 1 on failure, 2 on missing credentials

instructions/             # Agent prompts — edit these to change agent behaviour
  discover-examples.md
  create-example.md
  review-example.md
  fix-example.md

.github/
  workflows/              # CI workflows (tests, agents, auto-merge)
  ISSUE_TEMPLATE/         # Queue new examples or report broken ones

Numbering convention

Examples are numbered globally in increments of 10: 010, 020, 030 … The number appears as the directory name prefix. Gaps are intentional — they allow future insertions without renumbering.

Language support

Language	Test workflow	Marker file
Node.js / TypeScript	test-node.yml	package.json
Python	test-python.yml	requirements.txt or pyproject.toml
Go	test-go.yml	go.mod
Rust	test-rust.yml	Cargo.toml
.NET / C#	test-dotnet.yml	*.csproj

Setup

1. Add ANTHROPIC_API_KEY as a repository secret — required for all agent workflows
2. Add DEEPGRAM_API_KEY as a repository secret — required for E2E tests
3. Add partner credentials as needed (each example's .env.example lists them)
4. Enable auto-merge in repository Settings → General → Pull Requests
5. Configure branch protection on main to require status checks before merging

GITHUB_TOKEN is provided automatically by GitHub Actions.

Running agents locally

# Requires: ANTHROPIC_API_KEY set, gh auth login done, git configured
claude --model claude-opus-4-6 -p "$(cat instructions/discover-examples.md)"
claude --model claude-opus-4-6 -p "$(cat instructions/create-example.md)"