Imported from jeremylongshore/tons-of-skills-marketplace (skills/.curated/fathom-sdk-patterns/SKILL.md). Install upstream with npx skills add jeremylongshore/tons-of-skills-marketplace --skill fathom-sdk-patterns. Copyright stays with the author (MIT).

Fathom SDK Patterns

Overview

Build a narrow Fathom client boundary that makes consent, data handling, idempotency, redaction, and human ownership explicit.

Prerequisites

• Scoped development credentials, an approved recording/data policy, synthetic fixtures, and existing test conventions.

Instructions

1. Inject credentials only from the approved secret manager and validate request/response contracts at the boundary.
2. Use stable IDs, bounded retries, and redacted telemetry for sync/follow-up actions.
3. Unit-test with mocks and run one synthetic development integration check before promotion.

Output

• A test-backed SDK boundary with scoped credentials, safe errors, redacted observability, and rollback-friendly behavior.

Examples

Instantiate the client with a development secret reference, send a synthetic meeting/action event with an idempotency key, and assert the mocked CRM-sync payload. Record only opaque IDs/results; never log a recording, transcript, contact, or token.

Python Client

import os, requests
from dataclasses import dataclass
from typing import Optional

@dataclass
class FathomConfig:
    api_key: str
    base_url: str = "https://api.fathom.ai/external/v1"
    timeout: int = 30

class FathomClient:
    def __init__(self, config: Optional[FathomConfig] = None):
        self.config = config or FathomConfig(api_key=os.environ["FATHOM_API_KEY"])
        self.session = requests.Session()
        self.session.headers.update({"X-Api-Key": self.config.api_key})

    def list_meetings(self, limit: int = 20, **filters) -> list[dict]:
        params = {"limit": limit, **filters}
        resp = self.session.get(f"{self.config.base_url}/meetings", params=params, timeout=self.config.timeout)
        resp.raise_for_status()
        return resp.json().get("meetings", [])

    def get_transcript(self, recording_id: str) -> dict:
        resp = self.session.get(f"{self.config.base_url}/recordings/{recording_id}/transcript", timeout=self.config.timeout)
        resp.raise_for_status()
        return resp.json()

    def get_summary(self, recording_id: str) -> dict:
        resp = self.session.get(f"{self.config.base_url}/recordings/{recording_id}/summary", timeout=self.config.timeout)
        resp.raise_for_status()
        return resp.json()

TypeScript Client

class FathomClient {
  private apiKey: string;
  private baseUrl: string;

  constructor(apiKey?: string) {
    this.apiKey = apiKey ?? process.env.FATHOM_API_KEY!;
    this.baseUrl = "https://api.fathom.ai/external/v1";
  }

  private async get<T>(path: string, params?: Record<string, string>): Promise<T> {
    const url = new URL(`${this.baseUrl}${path}`);
    if (params) Object.entries(params).forEach(([k, v]) => url.searchParams.set(k, v));
    const resp = await fetch(url, { headers: { "X-Api-Key": this.apiKey } });
    if (!resp.ok) throw new Error(`Fathom ${resp.status}: ${await resp.text()}`);
    return resp.json();
  }

  async listMeetings(limit = 20) { return this.get<{meetings: any[]}>("/meetings", {limit: String(limit)}); }
  async getTranscript(id: string) { return this.get(`/recordings/${id}/transcript`); }
  async getSummary(id: string) { return this.get(`/recordings/${id}/summary`); }
}

Error Handling

Resources

• Fathom API Reference

Next Steps

Apply in fathom-core-workflow-a for meeting analytics.