Skip to main content

the bench, in the open

Follow the build log

There is no chat server right now, so the bench works in public: one email when something ships or breaks, plus the channels below. Free, low-noise, unsubscribe anytime.

$ crontab -l

Weekly rituals

Two standing entries in the bench crontab. Both run on the chat server, so they are on hold while it is closed.

Office hours

$ 0 18 * * FRI  # office hours, voice, 45–60m

Open voice channel. Bring a stuck automation, a homelab mystery, or a half-built agent. We triage live, screen-share the fix, and write down the runbook.

Build Log Friday

$ 0 9 * * FRI   # build-log, one message, one week

One post: what you shipped, what broke, what you learned. A benchmark or a postmortem beats a screenshot. Small progress counts — that is the whole point.

$ cat CONDUCT.md

How it works

Four guardrails keep the bench welcoming. Read them once; they are short on purpose.

Be kind, be specific

Assume good faith. Answer the question that was asked, then the one they meant.

Show your work

Paste the error, the command, the config. Receipts get better answers than vibes.

No secrets in chat

Never paste API keys, tokens, .env files, or client data. Redact first — always.

Search before you shout

The channel history and the Lab Notes are a decent first responder. Skim, then ask.

$ ls ~/receipts

No sign-up needed. The receipts are public — read the runbooks and browse what is on the bench.