VibeSignal

Controls a Lovense toy (via buttplug.io) from Mastodon interactions on your own posts: favourites, boosts, quote posts, and replies (which can carry custom intensity, duration, and pattern commands). Bundles the buttplug server (intiface-engine) itself — it's downloaded automatically on first run, no separate Intiface Central install needed. Includes a local web dashboard — the control room — for arming, firing test signals, and watching it all happen live.

Overview

• Mastodon-driven playback. Favourites, boosts, and quote posts each map to their own configurable intensity/duration/pattern; replies are freeform ("🌊 70% 12s") and get parsed.
• Armed/disarmed by design. Nothing reaches the toy until you flip the dashboard's arm switch — everything else (including a panic-stop) still logs, it just doesn't fire.
• Rate limiting. A configurable cooldown (an action's own duration plus a buffer you set from the dashboard) stops triggers from stacking faster than the toy can keep up.
• Live queue view. See what's currently playing and what's queued behind it, and who (or what event) triggered each one.
• Toy connection status + rescan. The dashboard shows which device is paired and whether it's currently connected, with a button to re-scan for devices without restarting the app.
• Scoping. React to every post, or lock onto a single specific post.
• History & leaderboard. Every interaction (delivered or not, e.g. while disarmed or rate-limited) is logged to SQLite, with a leaderboard of who's fired the most.

Setup

1. Create a Mastodon access token. On your instance, go to Preferences -> Development -> New Application, give it any name, and leave the scope at (at least) read. Create it, then copy the access token shown.

2. Install dependencies:

   python3 -m venv .venv
   source .venv/bin/activate
   pip install -r requirements.txt
   

3. Configure:

   cp config.example.yaml config.yaml
   

   Edit config.yaml: set mastodon_api_base_url and mastodon_access_token. Adjust the favourite_action / reblog_action / quote_action intensities and durations, and the max_intensity / max_duration_s safety ceiling, to taste.

4. macOS only — grant Bluetooth permission to your terminal app first. The bundled intiface-engine binary isn't code-signed, so macOS won't show its usual "wants to use Bluetooth" prompt — it just crashes on startup instead. Before running the app, go to System Settings -> Privacy & Security -> Bluetooth, click +, and add whatever terminal app you're running this from (Terminal, iTerm2, etc). If you skip this, python -m vibe_app.main will fail fast with an error telling you to do exactly this.

5. Run:

   python -m vibe_app.main
   

   The first run downloads intiface-engine into vibe_app/bin/. Once you see [toy] scanning for devices in the log, turn your Lovense toy on and put it in pairing/Bluetooth mode — that's the window in which it'll be discovered and grabbed (scan_timeout_s in config, default 15s). If nothing is found, just restart the app with the toy already on.

6. Open the dashboard at http://127.0.0.1:8420 — see The dashboard below.

How reactions map to the toy

Event	Behavior
Favourite	small fixed buzz (favourite_action in config)
Boost (reblog)	bigger/longer fixed buzz (reblog_action)
Quote post	its own bigger tier, ramps up (quote_action)
Reply	custom — see below

Replies are parsed for:

• Intensity: a number followed by %, e.g. 70%
• Duration: a number followed by s/sec/seconds, e.g. 10s
• Pattern emoji (first match wins): 🌊 wave · 💓/💗/💕 pulse · ⚡ burst · 🔥 escalate

So a reply like "🌊 70% 12s" plays a 12-second wave pattern peaking at 70% intensity. Anything not specified falls back to reply_default_intensity/reply_default_duration_s in config, and everything is clamped to max_intensity/max_duration_s regardless of what the reply asked for.

Plain @mentions that aren't actually replies to one of your posts are logged but don't trigger anything.

The dashboard

Open http://127.0.0.1:8420 — the app starts disarmed, so nothing reaches the toy until you arm it here.

• Arm / disarm — the master switch. While disarmed, everything still gets logged, just never sent to the toy.
• Panic stop — immediately halts the toy and clears anything queued.
• Toy status — shows which device is paired and whether it's currently connected, with a Rescan button to look for devices again without restarting the app (useful if the toy wasn't on yet, or dropped connection).
• Cooldown buffer — the rate limit: how many extra seconds (on top of an action's own duration) must pass before the next trigger is accepted. Editable live; persists across restarts.
• Manual trigger — fire a favourite/boost/quote-tier signal yourself, or a custom intensity/duration/pattern (or reply-style text like "70% 10s 🌊").
• Mode — react to all posts, or scope to one specific post (pick from your recent posts, or paste a URL).
• Queue — what's playing right now and what's queued behind it, with who (or what event) triggered each one.
• Top controllers — a leaderboard of who's fired the toy the most.
• History — the last 20 interactions, including ones that didn't fire (disarmed, rate limited, or over the queue cap), each showing intensity/duration/pattern and a link back to the triggering post where applicable.

Notes

• Everyone's interactions count — there's no allowlist by design (per your setup choice). The armed toggle and panic-stop button are your manual safety controls; the per-action safety ceiling in config is a hard limit on how much, applied regardless of who.
• max_queued_seconds caps how much total buzz time can be queued at once, so a post that goes unexpectedly viral can't queue an unbounded backlog — anything beyond the cap is still recorded in the dashboard history, just not sent to the toy.
• The rate-limit cooldown buffer seeds from rate_limit_buffer_s in config on first run only — after that, whatever you last set on the dashboard is what persists across restarts.
• History (including things that happened while disarmed) lives in history.sqlite3, gitignored alongside config.yaml (which holds your access token) and the downloaded engine binary.