Python Slack, Teams, and Webhook Notifier Module Prompt
Build a reusable Python notification module that sends messages to Slack, Teams, or generic webhooks with a unified interface, retries, rate limiting, and safe failure handling.
- Target user
- Engineers adding alerting/notifications to scripts and services
- Difficulty
- Intermediate
- Tools
- Claude, ChatGPT
The prompt
You are a senior Python engineer who builds notification plumbing that never crashes the job it is supposed to be reporting on. I will provide: - Which destinations I need (Slack, Microsoft Teams, generic webhook, email later) - Where credentials/webhook URLs live (env, secret manager) - Message types (info, success, warning, error) and any formatting needs Build a small library that: 1. **Exposes a unified interface** — a `Notifier` abstraction with `send(message, level, *, title, fields, link)` so callers don't care about the backend. Concrete adapters: `SlackNotifier`, `TeamsNotifier`, `WebhookNotifier`. Allow fan-out to multiple destinations. 2. **Formats per platform** — Slack Block Kit / attachments with severity colors; Teams MessageCard/Adaptive Card; generic JSON payload for webhooks. Map the abstract `level` to the right color/emoji per platform. 3. **Loads config safely** — webhook URLs/tokens from env or a secret manager, never hardcoded. Validate they exist at construction with a clear error. 4. **Retries transient failures** — exponential backoff with jitter on 429/5xx/network errors; honor `Retry-After`; cap attempts. Do not retry 4xx (bad payload/auth). 5. **Never raises into the caller by default** — notification failure should log loudly and return a result object, not crash the automation. Offer a strict mode for when a notification is mission-critical. 6. **Respects rate limits** — simple throttle so a loop firing 1000 alerts doesn't get the workspace rate-limited; optionally coalesce/batch. 7. **Redacts secrets** — never log the webhook URL or token; mask in any debug output. 8. **Is testable** — inject the HTTP client; provide tests that assert payload shape per platform using a mock transport. Output: (a) the base interface + adapters, (b) the retry/throttle helper, (c) config loading with validation, (d) `pytest` tests with mocked HTTP, (e) a usage example wiring it into a script's try/finally. Bias toward: a clean abstract interface, swallow-by-default-but-log errors, redacted secrets, and platform-correct formatting.