Skip to content

launch.md

Latest commit

 

History

History
59 lines (44 loc) · 2.28 KB

launch.md

File metadata and controls

59 lines (44 loc) · 2.28 KB

Launch materials — orb-async-dev

Launch tweet (draft, v1)

I built orb-async-dev: an OSS async coding agent that runs on your own Orb Cloud account.

Comment @orb fix the unicode login bug on any GitHub issue → a persistent sub-agent wakes in ~1s, runs a 10-gate verifier (build, tests, mutation check, self-review, red team…), and replies with a PR. Sleeps between LLM calls at ~$0/hr.

Devin clone. $0/mo infra idle. MIT, self-host in 60s.

github.com/AWLSEN/orb-async-dev

Alternate hook (shorter):

Text your repo. Come back to a PR.

orb-async-dev: self-hosted coding agent that wakes on webhook, runs a 10-gate verifier on itself, opens a PR. $0/hr idle. MIT.

github.com/AWLSEN/orb-async-dev

Demo gif checklist

Record at 1200×700, 15 fps, 20s max:

  1. iPhone mirrored to screen, GitHub Mobile open on a private demo issue
  2. Type @orb fix the unicode email 500 and send
  3. Cut to terminal showing [orchestrator] accepted task from @sam
  4. Cut to /logs page auto-refreshing with gate results streaming in
  5. End on the PR in the thread with the ## Verifier block visible

Export as docs/demo.gif and reference from the README hero.

Post-launch follow-ups (ordered by value)

  1. Telegram adapter (DM surface in addition to GH comments)
  2. Multi-repo support (one Orb computer, N connected repos)
  3. Dependabot-style continuous loop — wake on new dependency release webhook, open bump PR
  4. Public install metrics dashboard (opt-in via ORB_TELEMETRY=1)

Pilot-customer filter

Look for dev-twitter accounts retweeting the launch who:

  • Maintain ≥3 public repos AND
  • Tweet about dev tooling or infra in the last 30 days AND
  • Follow ≥1 of: @cognition_labs, @cursor_ai, @vercel, @railway

DM with a custom demo gif of a PR on their repo if the dogfood can cover it.

Failure modes to watch in the first 100 installs

  • auto-register hit rate on Orb (high = docs onboarding works; low = need clearer copy)
  • Ratio of hard-failed gates to opened PRs (> 30% hard = system prompt needs sharpening)
  • Cost watchdog trip rate (any trip = investigate runaway loops)
  • /logs 404 rate (indicates users setting logStore improperly)

Every one of these is already in the log lines we emit; grep your deploy's output file.