Keep /bunny working without Reddit API dependency

Smithy-bot commented

2026-09-19 21:48:57 +01:00

Member

Epic:
Story Points:

The /bunny command depends on random-bunny, which fetches anonymous Reddit JSON (https://reddit.com/r/{subreddit}/{sort}.json). Reddit is restricting that access, so the command is becoming unreliable.

Goal: keep /bunny returning a random rabbit image without relying on Reddit’s public/unauthenticated API.

Current behaviour

src/commands/bunny.ts picks a subreddit (rabbits, bunnieswithhats, buncomfortable, bunnytongues, dutchbunnymafia), calls randomBunny(subreddit, 'hot'), downloads the image, and embeds title / permalink / upvotes.
random-bunny (RabbitLabs package) has no OAuth; it only hits Reddit’s .json endpoints.

Options considered

Third-party bunny API (simplest) — e.g. Rabbit API (GET /api/random) or bnuy-api. Replace random-bunny in the bot. Loses subreddit metadata; depends on third-party uptime/catalog.
Self-host an image source (most durable) — run bnuy-api or a small image service on our infra. Full control; needs ops + image seeding/licensing.
Reddit OAuth in random-bunny (stopgap) — keep the same UX (subreddits, titles, scores) via authenticated Reddit API. Still Reddit-dependent (keys, rate limits, policy risk).
Stock image APIs — Unsplash / Pexels / Wikimedia with a rabbit query. Stable APIs but less community-bunny feel; keys + attribution required.

Recommendation from discussion: prefer option 1 or 2 for vylbot; treat OAuth in random-bunny as a delay, not a permanent fix. The command mainly needs a downloadable image URL.

Acceptance Criteria

GIVEN the bot is running without Reddit credentials or anonymous Reddit JSON access
WHEN a user runs /bunny
THEN they receive an embed with a rabbit/bunny image
AND the command does not call Reddit’s unauthenticated .json endpoints

GIVEN the image source is temporarily unavailable
WHEN a user runs /bunny
THEN the command fails gracefully with a clear error reply (no unhandled exception)

GIVEN a chosen non-Reddit source is adopted
WHEN the change is merged
THEN random-bunny is removed from vylbot-app dependencies if no longer used
AND tests for the bunny command are updated for the new source

Subtasks

Decide on source: third-party API vs self-hosted vs Reddit OAuth stopgap
Update src/commands/bunny.ts to use the chosen source
Adjust embed fields (title/footer/URL) for whatever metadata the new source provides
Update/remove random-bunny dependency as appropriate
Update bunny command tests
If self-hosting: stand up service + seed images + document ops
If changing random-bunny instead: track work in that repo and bump the package here

Notes

Related package: random-bunny @ RabbitLabs (also mirrored on Codeberg/npm)
Discord embed currently shows Reddit permalink and upvote count; those will need a replacement or simplification if leaving Reddit
Prefer not to scrape Reddit HTML as an alternative

Epic: Story Points: --- The `/bunny` command depends on `random-bunny`, which fetches anonymous Reddit JSON (`https://reddit.com/r/{subreddit}/{sort}.json`). Reddit is restricting that access, so the command is becoming unreliable. Goal: keep `/bunny` returning a random rabbit image without relying on Reddit’s public/unauthenticated API. ## Current behaviour - `src/commands/bunny.ts` picks a subreddit (`rabbits`, `bunnieswithhats`, `buncomfortable`, `bunnytongues`, `dutchbunnymafia`), calls `randomBunny(subreddit, 'hot')`, downloads the image, and embeds title / permalink / upvotes. - `random-bunny` (RabbitLabs package) has no OAuth; it only hits Reddit’s `.json` endpoints. ## Options considered 1. **Third-party bunny API (simplest)** — e.g. [Rabbit API](https://rabbit-api-two.vercel.app/docs) (`GET /api/random`) or [bnuy-api](https://git.scrunkly.cat/Michelle/bnuy-api). Replace `random-bunny` in the bot. Loses subreddit metadata; depends on third-party uptime/catalog. 2. **Self-host an image source (most durable)** — run bnuy-api or a small image service on our infra. Full control; needs ops + image seeding/licensing. 3. **Reddit OAuth in `random-bunny` (stopgap)** — keep the same UX (subreddits, titles, scores) via authenticated Reddit API. Still Reddit-dependent (keys, rate limits, policy risk). 4. **Stock image APIs** — Unsplash / Pexels / Wikimedia with a rabbit query. Stable APIs but less community-bunny feel; keys + attribution required. **Recommendation from discussion:** prefer option 1 or 2 for vylbot; treat OAuth in `random-bunny` as a delay, not a permanent fix. The command mainly needs a downloadable image URL. ## Acceptance Criteria GIVEN the bot is running without Reddit credentials or anonymous Reddit JSON access WHEN a user runs `/bunny` THEN they receive an embed with a rabbit/bunny image AND the command does not call Reddit’s unauthenticated `.json` endpoints GIVEN the image source is temporarily unavailable WHEN a user runs `/bunny` THEN the command fails gracefully with a clear error reply (no unhandled exception) GIVEN a chosen non-Reddit source is adopted WHEN the change is merged THEN `random-bunny` is removed from vylbot-app dependencies if no longer used AND tests for the bunny command are updated for the new source ## Subtasks - [ ] Decide on source: third-party API vs self-hosted vs Reddit OAuth stopgap - [ ] Update `src/commands/bunny.ts` to use the chosen source - [ ] Adjust embed fields (title/footer/URL) for whatever metadata the new source provides - [ ] Update/remove `random-bunny` dependency as appropriate - [ ] Update bunny command tests - [ ] If self-hosting: stand up service + seed images + document ops - [ ] If changing `random-bunny` instead: track work in that repo and bump the package here ## Notes - Related package: `random-bunny` @ RabbitLabs (also mirrored on Codeberg/npm) - Discord embed currently shows Reddit permalink and upvote count; those will need a replacement or simplification if leaving Reddit - Prefer not to scrape Reddit HTML as an alternative

Smithy-bot added the

labels

2026-09-19 21:48:57 +01:00

Vylpes added this to the 3.5.0 milestone

2026-09-19 21:50:29 +01:00

Vylpes modified the milestone from 3.5.0 to 3.4.0

2026-09-19 21:50:31 +01:00

Smithy-bot commented

2026-09-19 21:53:46 +01:00

Author

Member

Option 1 source recommendation

For a hosted third-party API (option 1), prefer Rabbit API over bnuy-api.

Rabbit API (recommended for option 1)

Endpoint: GET https://rabbit-api-two.vercel.app/api/random
Docs: https://rabbit-api-two.vercel.app/docs

Example response:

{
  "_id": "...",
  "breed": "hotot",
  "url": "https://firebasestorage.googleapis.com/...",
  "urlId": "4980879d"
}

Pros:
- Works today; returns a direct image URL (Firebase Storage)
- Not Reddit-sourced content — cleaner break from Reddit
- Simple JSON (breed, url, urlId) fits Discord embeds well
- Public docs / landing page
Cons:
- Small hobby project (Vercel + Firebase); no SLA — fine for option 1, not forever

bnuy-api (not preferred as the hosted default)

Endpoint: GET https://bnuy-api.scrunkly.cat/random

Example responses include Reddit-backed content, e.g.:

{
  "url": "https://bnuy-api.scrunkly.cat/images/....jpeg",
  "source": "bunnies",
  "original_url": "https://i.redd.it/....jpeg"
}

Pros:
- Self-hostable (compose.yml + MariaDB) — better fit if we move to option 2
- Has source / original_url for attribution
- Rate limiting exists
Cons for hosted option 1:
- Many images are still Reddit-derived (cached Reddit CDN originals)
- Only avoids Reddit’s API, not Reddit as the content pipeline
- Minimal public docs/landing; same single-operator reliability risk without cleaner metadata

Decision summary

Goal	Prefer
Hosted third-party now (option 1)	Rabbit API
Path to owning the source (option 2)	bnuy-api (self-host / fork)

Caveat: if Rabbit API disappears later, fall back to self-hosting (option 2).

## Option 1 source recommendation For a **hosted** third-party API (option 1), prefer **Rabbit API** over bnuy-api. ### Rabbit API (recommended for option 1) - Endpoint: `GET https://rabbit-api-two.vercel.app/api/random` - Docs: https://rabbit-api-two.vercel.app/docs - Example response: ```json { "_id": "...", "breed": "hotot", "url": "https://firebasestorage.googleapis.com/...", "urlId": "4980879d" } ``` - Pros: - Works today; returns a direct image URL (Firebase Storage) - Not Reddit-sourced content — cleaner break from Reddit - Simple JSON (`breed`, `url`, `urlId`) fits Discord embeds well - Public docs / landing page - Cons: - Small hobby project (Vercel + Firebase); no SLA — fine for option 1, not forever ### bnuy-api (not preferred as the hosted default) - Endpoint: `GET https://bnuy-api.scrunkly.cat/random` - Example responses include Reddit-backed content, e.g.: ```json { "url": "https://bnuy-api.scrunkly.cat/images/....jpeg", "source": "bunnies", "original_url": "https://i.redd.it/....jpeg" } ``` - Pros: - Self-hostable (`compose.yml` + MariaDB) — better fit if we move to **option 2** - Has `source` / `original_url` for attribution - Rate limiting exists - Cons for hosted option 1: - Many images are still Reddit-derived (cached Reddit CDN originals) - Only avoids Reddit’s *API*, not Reddit as the content pipeline - Minimal public docs/landing; same single-operator reliability risk without cleaner metadata ### Decision summary | Goal | Prefer | |------|--------| | Hosted third-party now (option 1) | **Rabbit API** | | Path to owning the source (option 2) | **bnuy-api** (self-host / fork) | Caveat: if Rabbit API disappears later, fall back to self-hosting (option 2).

Smithy-bot was assigned by Vylpes

2026-09-19 21:54:28 +01:00

Vylpes referenced this issue

2026-09-23 18:37:56 +01:00

Bunny command has stopped working #554

Vylpes added this to the 3.3 Sprint 9 project

2026-09-23 18:39:13 +01:00

Vylpes self-assigned this

2026-09-24 13:23:12 +01:00

Smithy-bot was unassigned by Vylpes

2026-09-24 13:23:12 +01:00

Vylpes added