~ $ cat ./mission.md
The center of the app is the engagement — a scoped, named container for a single piece of work. Pick a playbook or pick tools one-by-one; every scan output, finding, and screenshot auto-attaches and becomes a report. A Claude copilot watches the session and helps you interpret it. It suggests — it doesn't act.
>;)
The 75+ individual tools — discovery, recon, web exploit, AD, cloud, forensics — are the library that lives inside engagements, not the product itself.
A scoped container for the work: targets, exclusions, evidence, and a report at the end. Lab mode skips the rails; Engagement mode enforces them.
A Claude-powered copilot interprets output, proposes the next check, and drafts the report. Every active attack still waits for a human click.
Scan output, findings, screenshots — written to the engagement timeline as they happen. Append-only audit log records tool, target, argv, approver.
Bundles declare category, mode, and per-step rationale / success / approval. Passive Recon, Local Posture, Surface Inventory, Web App First Look ship built-in.
Mount, drag to /Applications. Right-click → Open on first launch.
↓ Download .dmgchmod +x, then run. Per-commit arm64 builds in Actions.
↓ Download .AppImageHeadless FastAPI sidecar. NET_RAW + NET_ADMIN for raw scans.
↓ Download docker-compose.ymlFirst launch prompts for a fresh Keychain entry the first time it touches a privileged tool (tcpdump, nmap SYN/UDP/OS). For the AI Assistant, open Settings → API keys and paste an Anthropic API key. All builds are unsigned — see docs/SIGNING.md for what code-signing would take.