Agent Console for Coding Agents

One local workspace to dispatch and monitor Codex and Claude Code sessions

Problem

Running several coding agents at once means jumping between terminals, hunting for the right session, checking progress, adding follow-up instructions and approving actions — attention gets shredded. Existing open-source attempts like agent-hub show this is a real, repetitive pain, but current solutions are personal scripts with no stable cross-agent session aggregation, status alerts or task routing.

Solution

A local desktop/web workspace that auto-discovers and connects sessions from different coding agents, groups them by project, shows live progress and pending confirmations, and lets you append instructions and read results from one place. Templated task dispatch plus completion notifications make one person running multiple agents feel like running a small team.

Users

Individual developers, indie hackers and small tech leads who run multiple AI coding agents in parallel

MVP

Auto-scan and connect Codex/Claude Code sessions; group task status and pending confirmations by project; append instructions and read results in one place; desktop notifications and retry alerts for failures; local-only storage, never upload code

Revenue

Free tier limited to 2 agent sessions; Pro subscription for unlimited sessions, multi-device sync and team sharing; optional one-time local license

Why now

Coding agents went from single-tool to multi-tool, and users are already writing glue scripts themselves — the productization window just opened, and an indie dev can win with workflow intuition