The ingestion engine
built for game studios

Collect, transform, and route player feedback from Steam, Discord, Twitch, Reddit, and more — all from a single 100MB binary that runs on your infrastructure.

100MB binary
21 connectors
No Java runtime, no bloat
terminal
$ curl -fsSL https://tsunagi.io/install | sh
Downloading tsunagi v0.8.2...
✓ Installed to /usr/local/bin/tsunagi
$ tsunagi server
Starting Tsunagi server...
✓ Server running on 0.0.0.0:8080
✓ 21 connectors loaded
✓ Studio Desktop ready
_
Scroll to explore
The Problem

Your community data is scattered.
Generic tools don't speak gaming.

Fragmented APIs

Different rate limits, pagination schemes, and formats across Steam, Discord, Reddit, Twitch. Each requires custom glue code.

Generic Data Tool Blind Spots

Standard tools don't understand Steam's deep historical backfill, Discord's real-time events, or Reddit's live streams. You build workarounds.

Data Leaves Your Network

SaaS solutions mean your player data, reviews, and community conversations leave your infrastructure. Compliance risk. Vendor lock-in.

The Solution

Collect. Transform. Route. Observe.
With full traceability.

1

Collect

21 gaming-native connectors. Steam, Discord, Twitch, Reddit, YouTube, Google Play, App Store...

rate limits pagination retries
2

Transform

Configuration-based processors. Mapper, filter, JavaScript, LLM enrichment, lang detection, removes duplicates, quality scoring.

YAML no code no duplicates
3

Route

Sinks to your stack. ClickHouse, Postgres, Kafka, S3/Parquet, Typesense, Temporal...

batching failed events
4

Observe

Full traceability with replay, shadow mode, and lineage. Debug any event through the entire pipeline.

trace replay shadow compare

Event Genealogy

Trace any event ID end-to-end through the pipeline

Replay

Re-process failed events after fixing the pipeline

Shadow Mode

Test new logic on live data without writing to sinks

Compare

Compare two runs side-by-side before promoting

Built on Experience

Battle-tested at
Lovelace.gg

Tsunagi powers Lovelace.gg, processing millions of community messages daily across game communities. When a connector fails, we trace the exact event, replay it, and fix the pipeline — without leaving our infrastructure.

Now you can use the same engine on your own servers.

1M+
Messages / day
50+
Game communities
99.9%
Uptime
0
Data leaves network
FAQ

Questions?
We've got answers.

Do I need Docker?
No. One binary, one config file. Download, run ./tsunagi server --config config.yaml. That's it. No Java runtime, no containers, no cloud signup.
Where does my data go?
Nowhere you don't control. Tsunagi reads from your APIs and writes to your databases. Your player data, reviews, and community conversations never leave your infrastructure. No SaaS, no vendor lock-in, no compliance headaches.
What happens when a connector breaks?
The Error Hub catches every failure with the full event payload. Fix your processor, click Replay, and the event re-processes. No data loss, no manual re-ingestion, no 3am panic.
Can I test changes without touching production?
Yes. Shadow Mode runs your new logic against live events without writing to your sinks. Validate, compare, then promote to production — zero risk, zero downtime.
How is this different from Airbyte or Fivetran?
Airbyte is generic SaaS built for Salesforce and Stripe. Fivetran is cloud-only with your data on their servers. Tsunagi is gaming-native (Steam cursors, Discord gateways, Twitch IRC), runs on-premise as a single binary, and has features like Replay and Shadow Mode that generic tools don't build.
What connectors are available?
21 gaming-native connectors: Steam (reviews, builds, community), Discord (bot, history), Twitch (API, chat, audio, control), Reddit (live, history), YouTube, App Store, Google Play, RSS, SQL, Kafka, S3, Webhook, and more. New connectors ship regularly.

Ready to own your
community data?

Get early access to Tsunagi. Be among the first game studios to run the ingestion engine that powers Lovelace.gg.

Contact us for enterprise licensing with SSO, clustering, and priority support.