### Install Dependencies and Run Example

Source: https://github.com/deepgram/recipes/blob/main/recipes/javascript/speech-to-text/v1/keyterm/README.md

Install project dependencies using npm and then execute the example script.

```bash
npm install
node example.js
```

--------------------------------

### Run Intent Recognition Example

Source: https://github.com/deepgram/recipes/blob/main/recipes/go/audio-intelligence/v1/intents/README.md

Execute the Go example to demonstrate intent recognition. Ensure the DEEPGRAM_API_KEY environment variable is set and dependencies are installed.

```bash
go run example.go
```

--------------------------------

### Fetch Specific SDK Example File

Source: https://github.com/deepgram/recipes/blob/main/instructions/generate-examples.md

Retrieves the content of a specific example file from an SDK's examples directory. This helps in understanding the correct import paths and client initialization patterns.

```bash
gh api "repos/{repo}/contents/examples/{filename}" --jq '.content' | base64 -d
```

--------------------------------

### Example Output with Numerals

Source: https://github.com/deepgram/recipes/blob/main/recipes/go/speech-to-text/v1/numerals/README.md

Shows the transcript output when the numerals feature is enabled. No specific setup is required beyond enabling the parameter.

```text
Transcript (with numerals): That's 1 small step for man, 1 giant leap for mankind.
```

--------------------------------

### Run Java TTS Encoding Example

Source: https://github.com/deepgram/recipes/blob/main/recipes/java/text-to-speech/v1/select-encoding/README.md

Commands to navigate to the project directory and execute the Java example using Maven.

```bash
cd recipes/java/text-to-speech/v1/select-encoding
mvn exec:java -Dexec.mainClass="Example"
```

--------------------------------

### Run Java Text-to-Speech Example

Source: https://github.com/deepgram/recipes/blob/main/recipes/java/text-to-speech/v1/select-model/README.md

Execute the Java example to demonstrate text-to-speech functionality with model selection. Ensure prerequisites are met before running.

```bash
cd recipes/java/text-to-speech/v1/select-model
mvn exec:java -Dexec.mainClass="Example"
```

--------------------------------

### Run Sentiment Analysis Example

Source: https://github.com/deepgram/recipes/blob/main/recipes/javascript/audio-intelligence/v1/sentiment/README.md

Execute the sentiment analysis script using Node.js after setting the API key and installing dependencies.

```bash
node example.js
```

--------------------------------

### Check SDK Examples Directory

Source: https://github.com/deepgram/recipes/blob/main/instructions/generate-examples.md

Checks if an 'examples' directory exists within the SDK repository and lists the names of files or directories within it.

```bash
gh api "repos/{repo}/contents/examples" 2>/dev/null | jq -r '.[].name'
```

--------------------------------

### Run Voice Agent Example

Source: https://github.com/deepgram/recipes/blob/main/recipes/java/voice-agents/v1/connect/README.md

Executes the Java voice agent example using Maven.

```bash
cd recipes/java/voice-agents/v1/connect
mvn exec:java -Dexec.mainClass="Example"
```

--------------------------------

### Run WebSocket TTS Example

Source: https://github.com/deepgram/recipes/blob/main/recipes/java/text-to-speech/v1/websocket-streaming/README.md

Commands to navigate to the project directory and execute the Java WebSocket TTS example.

```bash
cd recipes/java/text-to-speech/v1/websocket-streaming
mvn exec:java -Dexec.mainClass="Example"
```

--------------------------------

### Run Entity Detection Example

Source: https://github.com/deepgram/recipes/blob/main/recipes/java/audio-intelligence/v1/entities/README.md

Execute the Java example for entity detection. Ensure you have Java 11+ and Maven 3.9+ installed, and the DEEPGRAM_API_KEY environment variable is set.

```bash
cd recipes/java/audio-intelligence/v1/entities
mvn exec:java -Dexec.mainClass="Example"
```

--------------------------------

### Run the example

Source: https://github.com/deepgram/recipes/blob/main/recipes/python/text-to-speech/v1/generate-audio/README.md

Command to execute the Python script and generate the audio file.

```bash
python example.py
# Produces output.mp3
```

--------------------------------

### Run Voice Agent with Function Calling

Source: https://github.com/deepgram/recipes/blob/main/recipes/java/voice-agents/v1/function-calling/README.md

Navigate to the recipe directory and execute the Java example using Maven. Ensure you have Java 11+ and Maven 3.9+ installed, and the DEEPGRAM_API_KEY environment variable is set.

```bash
cd recipes/java/voice-agents/v1/function-calling
mvn exec:java -Dexec.mainClass="Example"
```

--------------------------------

### Find Existing Example Files (Java)

Source: https://github.com/deepgram/recipes/blob/main/instructions/pm.md

Specifically searches for Java recipes by looking for `src/main/java/Example.java` files within the recipe directory. This complements the standard example file search.

```bash
find "recipes/java/" -path "*/src/main/java/Example.java" 2>/dev/null | \
  sed "s|recipes/java/||" | sed "s|/src/main/java/Example\.java||" | sort
```

--------------------------------

### Run Java Streaming File Example

Source: https://github.com/deepgram/recipes/blob/main/recipes/java/speech-to-text/v1/streaming-file/README.md

Navigate to the example directory and run the Java code using Maven. Ensure the DEEPGRAM_API_KEY environment variable is set.

```bash
cd recipes/java/speech-to-text/v1/streaming-file
mvn exec:java -Dexec.mainClass="Example"
```

--------------------------------

### Intent Recognition Output Example

Source: https://github.com/deepgram/recipes/blob/main/recipes/cli/text-analysis/v1/intents/README.md

Example of the console output generated when running intent recognition.

```text
Intents:
  - Cancel subscription (0.92)
  - Request refund (0.88)
```

--------------------------------

### Find Existing Example Files (Standard Languages)

Source: https://github.com/deepgram/recipes/blob/main/instructions/pm.md

Searches for existing example files within a language's recipe directory. It excludes test files and module definition files. The output is a list of relative recipe paths.

```bash
find "recipes/{slug}/" -name "example.*" ! -name "*_test*" ! -name "*.mod" 2>/dev/null | \
  sed "s|recipes/{slug}/||" | sed "s|/example.*||" | sort
```

--------------------------------

### Run Audio Summarization Example

Source: https://github.com/deepgram/recipes/blob/main/recipes/java/audio-intelligence/v1/summarize/README.md

Execute the Java example to generate an audio summary. Ensure you have Java 11+, Maven 3.9+, and your DEEPGRAM_API_KEY environment variable set.

```bash
cd recipes/java/audio-intelligence/v1/summarize
mvn exec:java -Dexec.mainClass="Example"
```

--------------------------------

### Run Text-to-Speech v1 Example

Source: https://github.com/deepgram/recipes/blob/main/recipes/java/text-to-speech/v1/generate-audio/README.md

Execute the Java example to convert text to speech and save it to a file. Ensure prerequisites are met, including setting the DEEPGRAM_API_KEY environment variable.

```bash
cd recipes/java/text-to-speech/v1/generate-audio
mvn exec:java -Dexec.mainClass="Example"
```

--------------------------------

### Find Existing Example Files (Rust)

Source: https://github.com/deepgram/recipes/blob/main/instructions/pm.md

Specifically searches for Rust recipes by looking for `src/main.rs` files within the recipe directory. This complements the standard example file search.

```bash
find "recipes/rust/" -path "*/src/main.rs" 2>/dev/null | \
  sed "s|recipes/rust/||" | sed "s|/src/main\.rs||" | sort
```

--------------------------------

### Entity Detection Output Example

Source: https://github.com/deepgram/recipes/blob/main/recipes/java/audio-intelligence/v1/entities/README.md

This is an example of the expected output when entity detection is successfully run. It displays detected entities with their type labels and confidence percentages.

```text
PERSON: Neil Armstrong (95%)
ORGANIZATION: NASA (88%)
LOCATION: Moon (92%)
MISSION: Apollo 11 (85%)
```

--------------------------------

### Run Diarization Example in Java

Source: https://github.com/deepgram/recipes/blob/main/recipes/java/speech-to-text/v1/diarize/README.md

Execute the Java example using Maven to perform speaker diarization on an audio file. Ensure the DEEPGRAM_API_KEY environment variable is set.

```bash
mvn exec:java -Dexec.mainClass=Example
```

--------------------------------

### Example Output

Source: https://github.com/deepgram/recipes/blob/main/recipes/java/speech-to-text/v1/summarize/README.md

The expected format of the generated summary text.

```text
Summary: NASA astronauts conducted a spacewalk outside the International Space Station to perform maintenance and upgrades on critical systems.
```

--------------------------------

### Example Content Search Output

Source: https://github.com/deepgram/recipes/blob/main/recipes/rust/speech-to-text/v1/search/README.md

This is an example of the expected output when content search is enabled. It shows the 'search' array with terms, hits, start/end times, and confidence, along with the transcript and found term.

```json
Full Response:
{
  "search": [
    {
      "term": "NASA",
      "hits": [
        {
          "start": 5.2,
          "end": 5.7,
          "confidence": 0.95
        }
      ]
    }
  ],
  "channels": [...],
  ...
}

Transcript: So that was a big step for NASA.
Found search term 'NASA' at 5.20s
```

--------------------------------

### Run the .NET Example

Source: https://github.com/deepgram/recipes/blob/main/recipes/dotnet/audio-intelligence/v1/entities/README.md

Command to execute the .NET application after setting the required environment variables.

```bash
dotnet run
```

--------------------------------

### Sample Output Display

Source: https://github.com/deepgram/recipes/blob/main/recipes/javascript/text-to-speech/v1/bit-rate/README.md

Example of the console output indicating file size and configuration parameters.

```text
Saved output.mp3 (12345 bytes) — encoding: mp3, bit_rate: 48000
```

--------------------------------

### Run Sentiment Analysis Example

Source: https://github.com/deepgram/recipes/blob/main/recipes/python/audio-intelligence/v1/sentiment/README.md

Execute the Python script to perform sentiment analysis on audio. Ensure the DEEPGRAM_API_KEY environment variable is set and dependencies are installed.

```bash
python example.py
```

--------------------------------

### Example Utterance Response Structure

Source: https://github.com/deepgram/recipes/blob/main/recipes/dotnet/speech-to-text/v1/utterances/README.md

When `Utterances=true`, the response includes utterance data under `response.Results.Utterances`. Each utterance provides `Start`, `End`, `Transcript`, and optionally `Speaker` information.

```json
[0.5s - 3.2s] Yeah, as much as it's worth, I mean, you know.
[3.8s - 7.1s] From a scientific standpoint, it's fascinating.
[7.5s - 11.2s] The data we're collecting here is unprecedented.
```

--------------------------------

### Run Entity Detection Script

Source: https://github.com/deepgram/recipes/blob/main/recipes/cli/audio-intelligence/v1/entities/README.md

Execute this bash script to run the entity detection example. Ensure you have `curl` and `python3` installed and the `DEEPGRAM_API_KEY` environment variable set.

```bash
bash example.sh
```

--------------------------------

### Run Custom TTS Voice Agent Configuration

Source: https://github.com/deepgram/recipes/blob/main/recipes/java/voice-agents/v1/custom-tts/README.md

Execute the Java example to configure the voice agent with custom TTS settings. Ensure you have Java 11+ and Maven 3.9+ installed, and the DEEPGRAM_API_KEY environment variable is set.

```bash
cd recipes/java/voice-agents/v1/custom-tts
mvn exec:java -Dexec.mainClass="Example"
```

--------------------------------

### Run Text-to-Speech Stream Example

Source: https://github.com/deepgram/recipes/blob/main/recipes/java/text-to-speech/v1/stream-audio/README.md

Commands to navigate to the project directory and execute the Java application.

```bash
cd recipes/java/text-to-speech/v1/stream-audio
mvn exec:java -Dexec.mainClass="Example"
```

--------------------------------

### Install Deepgram CLI

Source: https://github.com/deepgram/recipes/blob/main/recipes/cli/speech-to-text/v1/detect-entities/README.md

Use this command to install the Deepgram CLI. Ensure you have curl installed.

```bash
curl -fsSL https://deepgram.com/install.sh | sh
```

--------------------------------

### View Example Output

Source: https://github.com/deepgram/recipes/blob/main/recipes/java/text-to-speech/v1/select-encoding/README.md

Shows the expected console output after successfully generating audio with a specified encoding.

```text
Audio saved: output.wav (96144 bytes, encoding: linear16)
```

--------------------------------

### Example Sentiment Analysis Output

Source: https://github.com/deepgram/recipes/blob/main/recipes/cli/text-analysis/v1/sentiment/README.md

This is an example of the output format when sentiment analysis is enabled.

```text
Sentiment: positive (0.93)
```

--------------------------------

### View Example Output

Source: https://github.com/deepgram/recipes/blob/main/recipes/go/text-to-speech/v1/stream-audio/README.md

Displays the expected console output when streaming audio successfully.

```text
Streamed audio: 12345 bytes
Content-Type: audio/wav
```

--------------------------------

### Format New SDK Review Summary

Source: https://github.com/deepgram/recipes/blob/main/instructions/engineer.md

Formats a summary for human review when a 'new-sdk' action is detected. This includes issue details and instructions for manual onboarding.

```text
=== New SDK Requires Human Review ===

Issue: #{number} — {title}
Created: {createdAt}

This issue was created because a new Deepgram SDK repository was discovered that is not
tracked in .deepgram/sdks.json.

To onboard this SDK:

1. Review the issue body for the repository name and language
2. Visit the GitHub repository and confirm it is an official Deepgram SDK
3. Edit .deepgram/sdks.json and add an entry:
   {
     "language": "{language}",
     "slug": "{language}",
     "repo": "{full_repo_name}",
     "manifest": "{appropriate manifest filename}"
   }
4. Create the recipes/{language}/ directory with an appropriate manifest file
5. Manually create a `test-{language}.yml` workflow — this CANNOT be done by an agent.
   Workflow files require the `workflows` GitHub token scope, which is not available to
   automated runs. The workflow also will not take effect until it is merged into main.
   A human must add this file via a separate PR.
6. Close the issue once onboarding is complete

No automated action has been taken. Human review is required.
===

```

--------------------------------

### Example Paragraph Output

Source: https://github.com/deepgram/recipes/blob/main/recipes/go/speech-to-text/v1/paragraphs/README.md

This is an example of the structured output when paragraph segmentation is enabled. Each paragraph is clearly delineated.

```text
Paragraphs:
Paragraph 1: That's one small step for man, one giant leap for mankind.
```

--------------------------------

### Example output

Source: https://github.com/deepgram/recipes/blob/main/recipes/python/text-to-speech/v1/generate-audio/README.md

The expected console output after successfully saving the audio file.

```text
Saved output.mp3: 48320 bytes (audio/mpeg)
```

--------------------------------

### Example Output of Audio Streaming

Source: https://github.com/deepgram/recipes/blob/main/recipes/cli/text-to-speech/v1/stream-audio/README.md

This is an example of the output you might see when audio is successfully streamed to stdout.

```text
Streamed 15432 bytes of audio to stdout
```

--------------------------------

### Example Transcript Output with Replacements

Source: https://github.com/deepgram/recipes/blob/main/recipes/go/speech-to-text/v1/replace/README.md

This is an example of how the transcript output will appear after the specified find and replace rules have been applied.

```text
Transcript (with replacements): That's one small step for person, one giant leap for humankind.
```

--------------------------------

### Example Text Summarization Output

Source: https://github.com/deepgram/recipes/blob/main/recipes/java/text-analysis/v1/summarize/README.md

This is an example of the output format for text summarization. It provides a concise summary of the input text.

```text
Summary: Deepgram provides AI-powered speech recognition and text-to-speech APIs with high accuracy, along with audio intelligence features accessible through multiple language SDKs.
```

--------------------------------

### Example Output of Key Term Prompting

Source: https://github.com/deepgram/recipes/blob/main/recipes/java/speech-to-text/v1/keyterm/README.md

This is an example of how the transcription output might look when using key term prompting.

```text
Yeah, and it's prior to them coming back in, they have to do a suit check on the ISS ...
```

--------------------------------

### Example Text Summarization Output

Source: https://github.com/deepgram/recipes/blob/main/recipes/go/text-analysis/v1/summarize/README.md

This is an example of the concise summary generated by the Deepgram text analysis API for the input text.

```text
Summary: Deepgram provides speech AI APIs including Nova-3 for transcription and Aura for text-to-speech, supporting real-time and pre-recorded audio across multiple languages.
```

--------------------------------

### Example Output of Utterances

Source: https://github.com/deepgram/recipes/blob/main/recipes/go/speech-to-text/v1/utterances/README.md

This is an example of the structured output provided by Deepgram's utterance detection, showing speaker turns and timestamps.

```text
Utterances:
Utterance 1 (Speaker 0): That's one small step for man, one giant leap for mankind.
  Start: 0.50s, End: 4.20s
```

--------------------------------

### Check Required Files in Sample Directory

Source: https://github.com/deepgram/recipes/blob/main/instructions/lead-reconcile.md

Iterates through each sample directory to verify the presence of an example file, a test file (with language-specific naming conventions), and a README.md file. Reports directories that are missing any of these required files.

```bash
# Check each directory
for dir in $(find recipes/ -mindepth 4 -maxdepth 4 -type d); do
  has_example=$(find "$dir" -maxdepth 1 -name "example.*" ! -name "*_test*" ! -name "*.mod" 2>/dev/null | head -1)
  # NOTE: .NET uses ExampleTest.cs (PascalCase), Java/Rust use subdirectories
  has_test=$(find "$dir" -maxdepth 1 \( -name "example_test.*" -o -name "example.test.*" -o -name "ExampleTest.*" -o -name "example_test.sh" \) 2>/dev/null | head -1)
  # For Java and Rust, test files are in subdirectories — always treat as having tests
  if echo "$dir" | grep -qE "/java/|/rust/"; then has_test="yes"; fi
  has_readme=$([ -f "$dir/README.md" ] && echo "yes" || echo "")

  if [ -z "$has_example" ] || [ -z "$has_test" ] || [ -z "$has_readme" ]; then
    echo "INCOMPLETE: $dir"
    [ -z "$has_example" ] && echo "  - Missing: example file"
    [ -z "$has_test" ] && echo "  - Missing: test file"
    [ -z "$has_readme" ] && echo "  - Missing: README.md"
  fi
done
```

--------------------------------

### Example Output of Key Term Prompting

Source: https://github.com/deepgram/recipes/blob/main/recipes/rust/speech-to-text/v1/keyterm/README.md

This is a sample output demonstrating how key term prompting can improve the recognition of specific phrases.

```text
NASA, that's one small step for man, one giant leap for mankind.
```

--------------------------------

### Example Profanity Filtered Transcript Output

Source: https://github.com/deepgram/recipes/blob/main/recipes/go/speech-to-text/v1/profanity-filter/README.md

This is an example of how a transcript appears when the profanity filter is enabled. Profane words are replaced with asterisks.

```text
Transcript (profanity filtered): That's one small step for man, one giant leap for mankind.
```

--------------------------------

### Run the Custom LLM Voice Agent Example

Source: https://github.com/deepgram/recipes/blob/main/recipes/java/voice-agents/v1/custom-llm/README.md

Executes the Java voice agent application configured for a custom LLM provider.

```bash
cd recipes/java/voice-agents/v1/custom-llm
mvn exec:java -Dexec.mainClass="Example"
```

--------------------------------

### Example Entity Detection Output

Source: https://github.com/deepgram/recipes/blob/main/recipes/cli/audio-intelligence/v1/entities/README.md

This is an example of the output format when named entity detection is enabled. It shows classified entities with their types.

```text
DATE: March 18th, 1965
PERSON: Alexei Leonov
ORG: NASA
```

--------------------------------

### List Instructions Directory

Source: https://github.com/deepgram/recipes/blob/main/instructions/vp.md

Lists the contents of the instructions directory. Use this to see available instruction files.

```bash
ls instructions/
```

--------------------------------

### Set Deepgram API Key and Install Dependencies

Source: https://github.com/deepgram/recipes/blob/main/recipes/javascript/speech-to-text/v1/filler-words/README.md

Set your Deepgram API key as an environment variable and install the necessary Node.js dependencies.

```bash
export DEEPGRAM_API_KEY="your-api-key"
npm install
```

--------------------------------

### Fetch SDK README

Source: https://github.com/deepgram/recipes/blob/main/instructions/generate-examples.md

Retrieves the README file for a specified SDK repository to get an overview of its API. The first 300 lines are usually sufficient.

```bash
gh api "repos/{repo}/readme" --jq '.content' | base64 -d | head -300
```

--------------------------------

### Example Output for Single-Channel Audio

Source: https://github.com/deepgram/recipes/blob/main/recipes/dotnet/speech-to-text/v1/multichannel/README.md

This is an example of the output when processing single-channel audio with Multichannel enabled. It shows a single channel with its corresponding transcript.

```text
Found 1 audio channel(s):

Channel 0: Yeah, as much as it's worth, I mean, you know...
```

--------------------------------

### Load SDK and Feature Configuration Files

Source: https://github.com/deepgram/recipes/blob/main/instructions/pm.md

Reads the `sdks.json` and `features.json` files from the repository root. `sdks.json` contains information about tracked SDKs, and `features.json` lists all expected recipe paths.

```bash
cat .deepgram/sdks.json
cat .deepgram/features.json
```

--------------------------------

### Example Sentiment Analysis Output

Source: https://github.com/deepgram/recipes/blob/main/recipes/dotnet/audio-intelligence/v1/sentiment/README.md

This is an example of the output format for sentiment analysis, showing the classification, text, and confidence score for each speech segment.

```text
Sentiment Analysis:
[positive] NASA announces a successful spacewalk mission where astronauts completed repairs.
  Confidence: 0.85

[neutral] The six-hour operation involved two crew members working outside the station.
  Confidence: 0.92

[positive] The mission restored full power generation capability to the space station.
  Confidence: 0.78
```

--------------------------------

### Check Existing Recipe Coverage

Source: https://github.com/deepgram/recipes/blob/main/instructions/generate-examples.md

Finds all existing example files for a given language to determine which recipes have already been generated. It excludes test files and `.mod` files.

```bash
find "recipes/{language}/" -name "example.*" ! -name "*_test*" ! -name "*.mod" 2>/dev/null | \
  sed "s|recipes/{language}/||" | sed "s|/example.*||" | sort
```