beam-ai-team/fathom-get-transcript

Fathom Get Transcript

Fetch the complete transcript for a specific Fathom meeting using its recording ID.

Safety Contract

Before fetching or saving a transcript, show the recording ID or URL, meeting title/date if known, output path, and whether speaker-attributed transcript text is included. Require explicit user approval in the current turn before fetching sensitive transcript content or writing files. Do not share transcripts externally from this skill.

Purpose

Retrieves the full speaker-attributed transcript from a Fathom recording. Use this when you have a meeting ID (from fathom-fetch-meetings or direct input) and need the complete conversation text for processing.

Quick Setup

Requirements: Fathom API key in .env

# Add to .env
FATHOM_API_KEY=your-api-key-here

Get your API key: Fathom Settings → API

Package Contents

fathom-get-transcript/
├── SKILL.md              # This file
└── scripts/
    └── get_transcript.py # Fetch transcript by recording ID

Workflow

Step 1: Get Recording ID

Accept recording ID from:

1. Output of fathom-fetch-meetings (recording_id field)
2. User-provided ID
3. Fathom meeting URL (extract ID from URL)

Enter recording ID (or paste Fathom URL):
> abc123-def456-789

Step 2: Fetch Transcript

python scripts/get_transcript.py --recording-id {ID}

API Call:

curl -s --request GET \
  --url 'https://api.fathom.ai/external/v1/recordings/{RECORDING_ID}/transcript' \
  --header 'X-Api-Key: {FATHOM_API_KEY}'

Step 3: Format Output

Raw Response:

{
  "transcript": [
    {
      "speaker": {
        "display_name": "John Smith",
        "matched_calendar_invitee_email": "[email protected]"
      },
      "text": "Let's start with the agenda...",
      "timestamp": "00:01:23"
    }
  ]
}

Formatted Output:

TRANSCRIPT: Meeting Title
Duration: 45 minutes
Speakers: John Smith, Jane Doe

---

[00:01:23] John Smith:
Let's start with the agenda...

[00:01:45] Jane Doe:
Thanks John. First item is the technical review...

Step 4: Return for Processing

Returns structured transcript data for downstream skills:

• Speaker-separated text blocks
• Timestamps for reference
• Raw text for analysis

Example Usage

# Fetch transcript by ID
python scripts/get_transcript.py --recording-id abc123-def456

# Fetch and save to file
python scripts/get_transcript.py --recording-id abc123 --output transcript.md

# Get JSON output for processing
python scripts/get_transcript.py --recording-id abc123 --json

Output Format

{
  "recording_id": "abc123-def456",
  "duration_seconds": 2700,
  "speakers": ["John Smith", "Jane Doe"],
  "transcript": [
    {
      "timestamp": "00:01:23",
      "speaker": "John Smith",
      "email": "[email protected]",
      "text": "Let's start with the agenda..."
    }
  ],
  "full_text": "John Smith: Let's start...\n\nJane Doe: Thanks John..."
}

Error Handling

| Error | Cause | Solution | |-------|-------|----------| | 401 Unauthorized | Invalid API key | Check FATHOM_API_KEY in .env | | 404 Not Found | Invalid recording ID | Verify ID from fathom-fetch-meetings | | 403 Forbidden | No access to recording | Ensure you have access in Fathom |

Related Skills

• fathom-fetch-meetings - Get recording IDs by filtering meetings
• create-meeting-minutes - Process transcript into formatted minutes
• process-client-meeting - Full meeting workflow

---

Version: 1.0 Integration: Fathom API Owner: Hassaan Ahmed