`go test -tags bench ./internal/tldr -run TestBench` runs the real /tldr
pipeline against one or more Ollama models and prompt variants on a chat dump
from disk (text copied from Discord, "Author: message" lines, or a
DiscordChatExporter JSON export) and writes a self-contained HTML dashboard
plus a JSON of the results. Scoring uses an optional facts file (a fact
passes only when one bullet holds all its details) and automatic checks:
numbers not in the chat, copying the chat, an English intro, cut-off output.
TestFactsSkeleton proposes candidate facts from the chat to prune by hand.
Chat dumps and results stay local: bench-out/ and the private folders are
git-ignored. See internal/tldr/tldrbench/README.md.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
/tldr [since] summarizes the messages in the channel it is run in that the
caller has not been summarized yet, using a small local model on Ollama, and
replies ephemerally. The answer streams into the reply as it is written.
- Per-user, per-channel cursor (tldr_cursors) so each run picks up where the
last left off; `since:` (e.g. 6h, 2d, max 7d) peeks without moving it.
First use covers the last 24h, and a stale cursor never reaches back more
than 7 days.
- Only summarizes channels both the caller and the bot can read.
- The transcript drops bots, system messages, reactions and filler, turns
links, GIFs, images and files into short tags ([link: site], [image],
[file: name], preview titles), marks replies, and writes a date marker
where the chat paused for 3h or more so the model can keep topics apart.
- The prompt asks for one bullet per conversation, oldest first, with a
worked example; it was tuned against a synthetic chat and a real export
with qwen3:4b-instruct-2507 (the default model).
- One model call at a time across the bot, one /tldr per user.
- docker-compose: Ollama service (CPU only, 8g memory cap: 6g killed the
runner mid-request), model pull on startup.
Needs the tldr_cursors migration (velox-dashboard-api 000016).
Co-Authored-By: Claude Opus 5.5 <[email protected]>
New AddToQueue appends to an already-active session without touching
voice (the dashboard can't join a channel on its own), and add_song's
dispatch case now branches: a voice_channel_id in the payload means
"start fresh", so it goes through EnqueueAndPlay (join + play) instead.
Query normalization (plain text -> YouTube search, stripping
autoplay/radio params off pasted YouTube URLs) moves out of /play's
handler into a shared NormalizeQuery, both entry points need it, not
just one.
Starting playback from the dashboard also posts the now-playing embed
into the voice channel's own built-in text chat, matching what /play
already does, instead of leaving it with nowhere to show up.
The dashboard has been writing skip/pause/volume/etc. requests into
music_commands with nothing on this end reading them. Adds a
dedicated LISTEN connection (can't use the pooled *sql.DB for this,
LISTEN has to stay bound to one specific connection) that wakes on
NOTIFY, drains unprocessed rows, and dispatches each to the matching
Manager function. A command's error is logged, not fatal, and every
row gets marked processed regardless of outcome so a permanently
broken command doesn't retry forever.
TrackEntry gets a QueueID generated once when a track is queued and
kept for its whole life there, and musicrepo.ReplaceQueue now upserts
by that ID instead of deleting and reinserting the whole queue on
every sync (which churned every track's database id constantly, even
ones that hadn't moved, and would've made remove/reorder commands
target the wrong track). RemoveFromQueue and Reorder are new Manager
functions built on top of that stable identity. Also adds repeat_song/
repeat_queue to the persisted now-playing state, and makes the two
existing toggle functions actually sync it, they never did before.
The queue and now-playing state only ever lived in the bot's memory,
so the dashboard had nothing real to show and a restart silently
wiped whatever was queued. Adds a musicrepo package and a syncState
call after every queue-changing action (play, skip, pause, volume,
stop, track end) that mirrors the current track and queue into
music_now_playing and music_queue.
No user-visible change yet, this is groundwork for the dashboard
queue/now-playing view.
Music could be started and controlled with no way to turn it off for
a server. Adds a music_settings table (mirrors the same one
velox-dashboard-api's dashboard toggle reads/writes), a /music toggle
command gated to Manage Server, and checks in /play, /queue, and
/volume so they refuse to run when a server has music turned off.
- Added PostgreSQL service to docker-compose.yml with health checks and persistent storage.
- Updated bot service to use environment variable substitution for database connection string.
- Upgraded Go version in Dockerfile to 1.26.1-alpine.
- Modified config loading to ignore missing .env file for better compatibility with Docker.
- Introduced Dockerfile for building the bot application using Go.
- Added .dockerignore to exclude unnecessary files from the Docker context.
- Updated docker-compose.yml to define the bot service and its dependencies, including environment variables for database and Lavalink configuration.
- Modified internal configuration to streamline environment variable handling.
- Added a logging service to manage server event logs including message edits, deletions, member joins/leaves, and moderation actions.
- Introduced commands to configure logging settings, including setting the log channel and enabling/disabling logging.
- Updated the README to document the new logging features and commands.
- Enhanced moderation commands to log actions taken on members.
- Introduced a new set of moderation commands including `/moderation purge`, `/moderation timeout`, `/moderation kick`, `/moderation ban`, `/moderation unban`, `/moderation untimeout`, and `/moderation slowmode`.
- Updated README to include details about the new moderation features and commands.
- Enhanced help command to display moderation tools and their usage.
- Introduced a new default role service and repository for managing default roles assigned to new members.
- Added command to set the default role through the bot's configuration.
- Implemented event handling to assign the default role to new guild members upon joining.
- Updated main application to include the new default role service in the bot's services.
- Introduced a new welcome service and repository for managing welcome messages, channels, and GIFs.
- Added commands to set welcome channel, message, DM, and GIF through the bot's configuration.
- Implemented event handling for new guild members to send welcome messages and DMs based on configuration.
- Updated main application to include the new welcome service in the bot's services.
- Added new music commands: play, queue, and volume.
- Implemented music management using disgolink for Lavalink integration.
- Updated bot initialization to include Lavalink host and password.
- Enhanced interaction handling for music commands, requiring DJ role for usage.
- Introduced now playing message with interactive buttons for controlling playback.
- Updated dependencies in go.mod for disgolink and snowflake.
FernandoJVideiraandClaude Opus 5.5