Skip to content

Commands

Global flags:

  • --provider none|openai|anthropic select AI provider for this run
  • --upgrade self-update from git checkout (see Upgrade section)

Run a deterministic system scan.

Flags:

  • --json output full JSON including remediation preview
  • --include-encrypted/--no-include-encrypted include heuristic search for encrypted-looking files
  • --verify/--no-verify run AI/heuristic verification to reduce false positives (default: verify)
  • --auto-approve auto-consent to safe, read-only probes (no prompts in TTY when set)
  • --max-probes N limit total verification probes (default: 5)

Examples:

Terminal window
cyberzard scan
cyberzard scan --json --include-encrypted
cyberzard scan --no-verify
cyberzard scan --auto-approve --max-probes 8
cyberzard --provider openai scan --include-encrypted

JSON output shape (abridged):

{
"scan": { /* scan results */ },
"remediation": { /* proposed plan (preview only) */ },
"verification": { // present when --verify
"success": true,
"verified_plan": { "total_actions": 2, "actions": [/* kept actions */] },
"dropped": [ { "action": { /* ... */ }, "reason": "..." } ],
"downgraded": [ { "action": { /* ... */ }, "reason": "..." } ],
"meta": { "probe_count": 3, "probes_skipped": 0, "consent_log": [/* categories + approvals */] }
}
}

Run a scan and print concise advice. Uses CYBERZARD_MODEL_PROVIDER (none|openai|anthropic); defaults to static advice.

Terminal window
CYBERZARD_MODEL_PROVIDER=none cyberzard advise
CYBERZARD_MODEL_PROVIDER=openai OPENAI_API_KEY=... cyberzard advise --include-encrypted
cyberzard --provider anthropic advise

Ask the minimal local agent to perform actions with tools.

Terminal window
cyberzard agent "scan the server"
cyberzard agent --steps 5 "read /etc/passwd"

Run cyberzard --help for full list.

CommandPurposeKey Options
scanRun all scanners, list findings--json, --verify/--no-verify, --auto-approve, --max-probes
adviseGenerate concise advice from scan--json, --include-encrypted
agentMinimal ReAct loop over safe tools--steps N, --show-plan
chatInteractive chat with permission-gated tools & SQLite history--verify/--no-verify, --auto-approve, --max-probes, --session
show-promptPrint the agent system prompt
versionShow version
upgradeSelf-update from git checkout
email-securityEmail stack scan + hardening preview / guided exec--json, --run, --dry-run/--no-dry-run, --max-risk
email-fixFull email remediation guide + optional execution--json, --run, --dry-run/--no-dry-run, --max-risk

A beautiful, permission-aware chat loop. Cyberzard stays focused on CyberPanel anomaly detection.

Flags:

  • --verify/--no-verify AI/heuristic verification of suggested actions (default: verify)
  • --auto-approve auto-consent to safe read-only probes
  • --max-probes N limit total probes during verification
  • --session ID conversation session id for persisted history

Examples:

Terminal window
cyberzard chat
cyberzard chat --no-verify
cyberzard chat --auto-approve --max-probes 8
cyberzard chat --session ops

In-chat commands:

  • /history [n] — show last n messages (default: 10)
  • /clear — clear current session history
  • /sessions — list available sessions
  • /switch <id> — switch active session

Cyberzard’s n8n setup and email troubleshooting flows now integrate with the AI agent for safer, permission‑gated execution and provide persistent logs + memory records:

  • n8n (cyberzard n8n-setup)

    • Flags: --interactive/--no-interactive, --auto-approve, --json-out
    • Execution logs: placed next to the applied script (e.g., n8n_setup_native.sh.log)
    • Memory: recorded under chat session n8n with applied/aborted status, mode, and paths
  • Email (email-security, email-fix)

    • Flags: --log-dir <path> (JSON log per run), plus existing --run, --dry-run/--no-dry-run, --max-risk, --auto-approve, --ai-refine/--no-ai-refine
    • Logs: a JSON file saved in the chosen directory (e.g., ./logs/email_exec_1694971234.json)
    • Memory: recorded under chat session email-troubleshooting with summary counters and log_path

In chat, you can list sessions with /sessions, switch with /switch n8n or /switch email-troubleshooting, and view history with /history.