Early access — free while we're in testing, pricing not decided yet

MULTIPLE SOURCES. ONE LOG.

Every source drains
into one log.

Commits, incidents, decisions, and AI summaries — everything that happens on a project, in one running log. Offline-first, synced everywhere.

Start loggingSee how it works
api-service

GitHubmerged fix/auth-refresh-race (#482)

Paged in @dana— redis latency spike, mitigated by scaling the read replica.

AI Summary

Shipped the auth fix, resolved the redis incident, kept the read-replica scaling policy from today's page.

Click to start typing…

FEATURES

Nothing to configure.
Nothing to lose.

AI Summary

AI that edits, not replaces

Summarize a busy day into one entry, or refine grammar and tone without losing your own words.

Paged in@dana— see[[ incidents ]]for context.

Mentions and references, inline

@mention a teammate, [[ link another drain or entry, or leave a comment anchored to the text.

synced offline — 3 pending

Offline-first, synced everywhere

Keep writing with no connection. Every device sees every entry the moment you're back online.

Encrypted client-side

Private by default

Private drains are sealed before they leave your device — not even the sync server can read them.

FOR AGENTS

A knowledge base your agents actually have.

Connect Claude Code, or any MCP-compatible agent, directly to a drain. It queries the log for context before it starts — why a decision was made, what broke last time — and writes its own entries as it works, attributed and indistinguishable in the timeline from anything a person typed. One durable memory, shared by every agent you run.

claude_desktop_config.json
{
  "mcpServers": {
    "drains": {
      "command": "npx",
      "args": ["-y", "github:drains-dev/mcp"]
    }
  }
}

queriedwhy was dynamo replaced with postgres for orders?

via mcpbumped retry backoff on the sync client, logged why

USE CASES

However you ship.

Side project

Just me

Solo builder

A personal engineering journal — why you built it that way, searchable months later.

paged 02:14mitigated 02:26resolved 03:02

On-call & SRE

Deploys, pages, and mitigations on one timeline — a clean handoff at the end of a shift.

DRMShared

Small teams

One drain per project — replaces the scattered Slack thread nobody can find later.

drains.dev/p/8fq2m

Build in public

Publish a read-only share link — a running changelog people can actually follow.

HOW IT WORKS

Three steps to a better log.

01

Create an identity

Your name and email. It becomes a deterministic color and ID that follows you everywhere — no password reset, no verification email.

02

Open a drain

One per project, incident, or team. Keep it private, share a link with collaborators, or publish it read-only.

03

Start typing

Every line is its own entry, saved as you type it. GitHub activity and agent updates show up in the same place.

getting started
01
Ada Lovelace
ada@example.com
02
+ New drain
Private
Shared
03
Deployed API v2.3.1

Stop losing context.

Your next incident review will thank you. Start logging in seconds — no account setup, no configuration.

Free while we're in testing — we haven't decided on pricing yet.