Allomos
Switch agents and teammates.
Keep the whole project on the record.
When one AI agent or account runs out, another picks up exactly where it stopped. Every task, its status and its handoff notes are written down in one place — your own repository — instead of scattered across chats.
Vibe coding leaves its history in a hundred chats
A project moves through many sessions: Claude Code, Codex, several accounts, several people. A quota can end a session without warning; the next one starts cold, and what was decided lives only in an old conversation. In one real multi-agent project:
- 185 KB of handoff notes that each new session re-read to resume;
- 6 places where one task's status was copied by hand;
- 3 sessions claiming tasks at once, settled by merge conflict;
- 0 answers to “when will it be done?”
Four things Allomos does
Agents and teammates
Hand the project from one AI agent or account to another, or to a teammate, and pick up exactly where it stopped. Nothing is lost at the switch.
The whole process, on the record
Every task, its status, handoff notes and decisions are recorded in one place, not spread across chats — easy to review, and easy to maintain for years.
Get better with your agents
Work with your agents in a way that improves over time: what each one does well, what to hand over, and a process that keeps you in control.
Where it stands, when it lands
Progress by area, what waits on you, and a forecast finish date drawn from your real pace.
Switching: Claude runs out, Codex picks up
Every agent uses the same alm commands. One step to hand over, one step to take over — between agents, between accounts, between people. The next session reads the task and its note, not a 185 KB history.
# Claude Code · account 1 · quota nearly used
alm label A01-03 "implemented"
alm release A01-03 --note "form done; tests next"
# Codex · account 2 · a minute later
alm next
A01-03 Sign-in page ready · note: tests next
alm claim A01-03Task management: nothing hides in a chat
Tasks live as readable files in .allomos/taskbag/, versioned by your own Git — not inside one agent's memory or one account. What the taskbag keeps:
- Every task and its state — ready, in progress, waiting, blocked, done — in one place, with one owner at a time.
- Proven steps — each task climbs a ladder (implemented → locally verified → deployed → checked by a person), so progress is shown, not claimed.
- Handoff notes and the AI's own decisions — what was done, what's next, and every routine choice the agent made, so you can review them later.
- A queue for you — what needs your check, your answer or your review, in one list.
Months later, when you or a new teammate come back to the code, the reason behind each change is still there.
AI skills In development
Working with agents is a skill, and it compounds. Allomos will help you learn what each agent is good at, what to pass on at a switch, and a simple way of working — one task per outcome, proven steps, your approval before anything leaves your machine — so every project goes better than the last.
Progress In development
Where is the project, and when will it land? Progress by area, the tasks waiting on you, and a burndown with a forecast finish date based on how fast the work really moves.
An example of the forecast to come, not real data.
Free on your machine. Pay only for the cloud.
Free
No account, no card, no trial clock. Your tasks stay on your machine and in your repository.
Optional
Live sharing, email invitations and roles for teams. Pricing to be announced.
Early development
v0.1, the macOS local tracker, is out. v0.2 (macOS and Windows, editing in the app) is in progress. Follow along or try it on GitHub.