Oneshot
A Slack-native work agent with proactive watchkeeping.
Andrew built Oneshot to remove the repetitive layer around his work and give the team a capable operator inside Slack. A request can begin in a DM, mention, watched thread, GitHub event, or bounded schedule; the agent carries it into the correct repository and returns material progress to the conversation where the work started.
It is designed as an operational reflection of Andrew rather than a generic chatbot: precise about ownership, sceptical of unverified state, able to draw on a deep catalogue of specialised skills, and expected to finish the whole authorised loop. Tests, review, pull requests, deployment, and live proof remain distinct gates.
Its proactive watches keep an eye on Andrew-relevant pull requests, release drift, critical journeys, repository hygiene, team handovers, and the health of adjacent agents. Quiet checks stay quiet. Work reaches Slack when there is a result, a real blocker, or a decision that still belongs to a person.
Operating model
What makes Oneshot itself.
- 01
Slack-native cockpit
Thread-scoped context, native progress, cancellable work, an App Home, and handovers that begin where the team already works.
- 02
Proactive watchkeeping
Event-driven and scheduled loops notice eligible PRs, incidents, drift, and unanswered responsibilities without turning every check into noise.
- 03
Skill-routed execution
A curated catalogue supplies the exact engineering, product, QA, operations, research, and communication procedure needed for each job.
- 04
Evidence-led delivery
Repository state, CI, deployment identity, live behaviour, and the authority to act are verified separately before work is called complete.
