`cybros ai review`

Run the 9-agent AI analysis on a scan and render each agent's verdict, the overall risk score, and the summary. Fix patches produced by the agents are surfaced on individual findings (see cybros findings show).

Synopsis

cybros ai review --scan SCAN_ID [OPTIONS]

Options

Flag	Default	Description
`--scan`	required	Scan id to run the AI analysis on.
`--timeout`	`600`	Max seconds to wait.
`--no-wait`	off	Trigger the analysis without waiting.

Plus the global flags: --json, --api-url, --no-color.

Examples

Run and wait

cybros ai review --scan scan_8a1b...

⠋ AI analysis (9/9 agents) — completed 0:01:12
╭─ AI analysis ────────────────────────────────╮
│ Status: completed                             │
│ Risk score: 58                                │
│ Confidence: 92%   Credits: 120                │
╰───────────────────────────────────────────────╯
                Agent verdicts
┏━━━┳━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ # ┃ Agent            ┃ Status    ┃ Confidence ┃
┡━━━╇━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ 1 │ triage           │ completed │        97% │
│ 2 │ false-positive   │ completed │        94% │
│ 9 │ remediation-plan │ completed │        88% │
└───┴──────────────────┴───────────┴────────────┘

Trigger without waiting

cybros ai review --scan scan_8a1b... --no-wait

AI analysis ai_5c2d… queued.

JSON

cybros --json ai review --scan scan_8a1b... | jq '{
  risk: .risk_score,
  confidence: .overall_confidence,
  agents: (.agent_runs | length)
}'

Notes

Triggering is idempotent-ish: if an analysis already exists for the scan, the server returns the current run rather than starting a duplicate.
The command polls GET /scans/{id}/ai-analysis until the analysis reaches a terminal state (completed, failed, cancelled) or --timeout elapses.

Exit codes

Code	When
`0`	Analysis completed / queued.
`1`	Unknown scan, timeout, or API error.
`2`	Not authenticated.

cybros ai review