◈ MISSION CONTROL · GROUND CONTROL FOR AI AGENTS ◈

AGENT
COCKPIT

REAL-TIME COMMAND CENTER FOR AI AGENTS

▶ LAUNCH MISSIONVIEW ON GITHUBVIEW DOCS →
SYSTEMS NOMINAL · v0.1 · LOCAL-FIRST

◈ FEATURE STATUS

FULL-SPECTRUM AGENT OVERSIGHT

SESSION MANAGER

MULTI-AGENT CONTROL

Launch and monitor multiple Claude Code and Codex sessions simultaneously. Every agent has its own terminal, timeline, and approval queue — unified in one control room.

▸ 12 AGENTS ACTIVE
APPROVAL QUEUE

HUMAN IN THE LOOP

Every destructive command surfaces for review before execution. Risk-rated in real time — critical, high, medium, low — so you stay in control without reading every log.

▸ 04 PENDING
MEMORY PANEL

PERSISTENT CONTEXT

Agents remember what they've built. Memory files persist across sessions and surface in the UI so you know exactly what each agent carries into the next task.

▸ 08 MEMORY FILES
TIMELINE VIEW

FULL AUDIT TRAIL

Every tool call, file write, and approval decision is logged with timestamps. Replay any session step-by-step to understand exactly what your agents did and why.

▸ 247 EVENTS LOGGED
LIVE TERMINAL

DIRECT CHANNEL

A real terminal window for each agent, streaming output live. Send messages, interrupt tasks, or inject context mid-run without leaving the cockpit.

▸ 3ms LATENCY
DIFF VIEWER

CHANGE REVIEW

Before any file is written, see the full diff inline — additions, deletions, context lines. Approve, reject, or edit before the agent commits the change.

▸ 14 FILES CHANGED

◈ APPROVAL QUEUE

RISK-RATED COMMAND REVIEW

AGENT SESSION · wizardly-mclaren · /projects/monorepo
⬢ CRITICAL
BASH rm -rf .next node_modules && npm run build

Deletes build artifacts and reinstalls all dependencies. Destructive, irreversible without re-install.

▲ MEDIUM
WRITE src/api/auth.ts

Rewrites authentication handler. Updates token validation logic and session expiry behaviour.

▼ LOW
READ package.json

Reading project manifest to inspect current dependency versions before suggesting upgrades.

● LIVEWAITING FOR APPROVAL · 3 PENDING14ms

◈ AGENT ROSTER

MEET YOUR CREW

ASTRONAUT
ASTRONAUTCODING
NINJA
NINJAPLANNING
ROBOT
ROBOTTESTING
ALIEN
ALIENREADING
PIRATE
PIRATEIDLE
GHOST
GHOSTBLOCKED
HOLOGRAM
HOLOGRAMDONE
CAVEMAN
CAVEMANPLANNING
MED-WOMAN
MED-WOMANCODING
MONKEY
MONKEYTESTING

◈ MISSION BRIEFING

THREE STEPS TO GROUND CONTROL

STEP 01

LAUNCH SESSION

Run npx @agentcockpit/agent-cockpit in your project directory. The cockpit opens at localhost:54321, ready to spawn agents.

STEP 02

MONITOR & CHAT

Watch every agent's terminal in real time. Send instructions, inspect the timeline, and review diffs from a single unified interface.

STEP 03

REVIEW & APPROVE

Risky commands pause for your review. Approve, deny, or edit before execution. Full audit trail kept for every session.

◈ LAUNCH SEQUENCE

GET AIRBORNE IN 60 SECONDS

TERMINAL · BASH
PREREQUISITES
Claude Code
macOS / Linux
$curl -fsSL https://claude.ai/install.sh | bash
Windows
>irm https://claude.ai/install.ps1 | iex
Codex (optional)
All platforms
$npm install -g @openai/codex
RUN AGENT COCKPIT
$npx @agentcockpit/agent-cockpit

Opens at http://localhost:54321. Requires Node.js 22+.
→ Full installation guide