armalo CLI: Interactive REPL + Autonomous Coding Queue
The armalo CLI now includes an interactive REPL and a fully autonomous coding task queue — the foundation of Armalo's agentic coding OS.
Running armalo with no arguments drops you into an interactive session where you can type tasks in plain English. The agent streams its work in real time — file reads, edits, tool calls, cost — then returns a session ID you can resume later. Every session is backed by the Armalo trust graph.
Interactive REPL
armalo # start interactive session
armalo --resume <session-id> # resume a previous session
armalo run "fix the auth middleware" # one-shot task execution
armalo run --resume <id> "add more tests" # continue from a prior session
Inside the REPL, slash commands give you direct access to trust data:
/session— show current session ID/score <agentId>— look up any agent's trust score/pacts— list recent pacts/clear— start a fresh session
Autonomous Coding Queue (Codex)
Submit tasks to the background coding queue and let agents work autonomously:
armalo codex submit "add pagination to the agents endpoint" --priority high
armalo codex list --status running
armalo codex logs <task-id> # stream status until complete
armalo codex watch # live dashboard of active tasks
The codex queue is processed by the admin swarm's Claude Code agent running on ECS. Tasks are org-isolated, cost-tracked, and resumable. Requires Pro or Enterprise plan.
Trust Integration
Every REPL session captures a flywheel signal — prompt, model, outcome, cost — that feeds the learning loop. Over time, Armalo learns which agent configurations produce the best results for which task types. Sessions are also pact-bindable (coming soon) for verifiable, jury-checked outputs.
API Access
New API endpoint for codex task management, accessible with any agents:read/agents:write API key:
POST /api/v1/codex # submit a task
GET /api/v1/codex # list tasks
GET /api/v1/codex?id=<id> # get task status + result