Planner and coder missions
A planner turns a goal into scoped work, then coder agents implement and report back through the agent bus.
A planner agent runs coder agents on scoped missions, in parallel, using Claude Code, Codex CLI, and OpenAI-compatible providers you control.
ZecruAI is free. You bring your own subscriptions or API keys. We add zero provider markup, so you are not paying a repriced middle layer for model access.
Unsupported platform or offline release lookup; GitHub Releases stays available.
Split the work into parallel missions, put each agent on a branch, then review and merge changes from the app.
Agents do their work in separate worktrees so parallel edits stay reviewable.
Durable project facts and mission summaries keep later turns grounded.
agent bus
planner -> coder: build the getting-started page
coder -> planner: tests green, branch pushed
Why it exists
ZecruAI keeps agents, project folders, provider credentials, and local config on your machine while giving them enough protocol to work as a team.
A planner turns a goal into scoped work, then coder agents implement and report back through the agent bus.
Every agent works on its own branch. Review changes and merge from the app without mixing parallel work in one folder.
Guided setup gets you from download to first agent in minutes: CLI checks, auth, project folder, and first prompt.
Durable project memory and mission distillation carry decisions, constraints, and learned context across long sessions.
Works with Claude Code, Codex CLI, and OpenAI-compatible bring-your-own providers.
Project mappings and settings are stored locally with atomic writes and backup recovery for daily use.
Pricing advantage
Editor-subscription tools often resell model access through per-seat or per-token pricing. ZecruAI is the coordination layer for the subscriptions and keys you already have.
The desktop orchestrator is free today.
Claude Code, Codex CLI, or BYO OpenAI-compatible API keys.
Run multiple agents without paying ZecruAI a model-access markup.
A planner turns a goal into an executable mission.
Coder agents receive scoped work on isolated branches and report progress back through the bus.
Memory, repo maps, and mission summaries keep the next turn grounded.
Screenshots
These slots intentionally show exact file paths. Replace them with real captures from the desktop app before launch.
Mission canvas
apps/website/public/screens/canvas.pngDrop in the desktop mission/canvas view.
Agent chat
apps/website/public/screens/agent-chat.pngDrop in an active planner/coder chat with usage visible.
Project memory
apps/website/public/screens/project-memory.pngDrop in memory/distillation UI once captured.
ZecruAI coordinates agents. It does not resell model access, add analytics, or set cookies in this static site.