Skip to main content
Fight Mode is Hiro’s continuous threat hunting capability. Unlike the Chat interface which responds to individual queries, Fight Mode runs in a loop—continuously scanning your integrations, discovering new indicators, and proposing remediation actions in real-time.

When to Use Fight Mode

Use Fight Mode when:
  • Active incident — You’re responding to a confirmed or suspected breach
  • Threat hunting — You want to continuously monitor for specific indicators
  • Incident response — You need real-time visibility and rapid remediation
  • Investigation at scale — The threat involves multiple systems and indicators
Fight Mode is designed for situations where threats evolve faster than you can manually query each system.

How Fight Mode Works

Fight Mode operates in continuous cycles, scanning your integrations and reporting findings in real-time.

Response Times

PhaseExpected Time
Initial scan after session start5–10 seconds
Each scanning cycle~15 seconds
Action execution after approval1–3 seconds
New indicator discovery scan5–10 seconds
Response times depend on the number of connected integrations and the volume of activity being analyzed. Complex queries across multiple systems may take longer.

The Scanning Loop

Each cycle, Hiro:
  1. Queries all connected integrations for activity matching known threat indicators
  2. Discovers new indicators — IPs, users, credentials, resources
  3. Reports specific findings with exact timestamps from logs
  4. Proposes remediation actions when threats are confirmed
  5. Executes approved actions immediately
  6. Adds new indicators to the threat context for future scans

Starting a Fight Mode Session

1

Navigate to Fight Mode

Click Fight Mode in the left sidebar.
2

Start a new session

Click New Session to begin.
3

Provide initial indicators

Enter the initial threat context:
Investigate suspicious login from IP 185.220.101.1. The user
john@company.com reported their account was compromised.
4

Watch Hiro hunt

Fight Mode begins scanning immediately. You’ll see real-time updates as Hiro discovers activity.

The Fight Mode Interface

The Fight Mode interface is divided into several panels:

Threat Indicators Panel

Shows all tracked indicators with their provenance:
IndicatorTypeSource
185.220.101.1IP AddressInitial input
john@company.comEmailInitial input
AKIA...XYZAWS Access KeyDiscovered from CloudTrail
jane@company.comEmailDiscovered from shared session

Findings Feed

Real-time stream of discoveries with timestamps:
[14:32:15] Okta: User john@company.com logged in from 185.220.101.1
[14:32:16] Okta: Session created, app access granted to 3 applications
[14:32:18] AWS: CloudTrail shows API calls from john@company.com
[14:32:19] AWS: New access key created: AKIA...XYZ
[14:32:21] GitHub: Repository clone from john@company.com

Proposed Actions Panel

Actions awaiting your approval:
PROPOSED ACTION: Suspend Okta User
─────────────────────────────────
User: john@company.com
Reason: Compromised credentials detected
Evidence:
  - Login from known malicious IP 185.220.101.1
  - Unusual access key creation in AWS
  - Repository access outside normal hours
Confidence: High (92%)

[Approve]  [Reject]  [More Info]

Executed Actions Log

Audit trail of all actions taken:
[14:35:22] EXECUTED: Cleared Okta sessions for john@company.com
[14:35:24] EXECUTED: Deactivated AWS access key AKIA...XYZ
[14:38:15] EXECUTED: Suspended Okta user john@company.com

Threat Indicator Types

Hiro tracks multiple indicator types during Fight Mode:
TypeExampleUse Case
IP Address185.220.101.1Source of malicious activity
User Emailjohn@company.comCompromised or suspicious user
AWS Principal ARNarn:aws:iam::123:user/adminCloud identity
AWS Access KeyAKIA...Potentially stolen credentials
Okta User ID00u1a2b3c4d5e6f7gInternal Okta identifier
GitHub UsernamejsmithSource code access
Custom IdentifierAny stringLLM-discovered identifiers
When a new indicator is discovered, Hiro automatically:
  • Adds it to the tracking list with source information
  • Scans the last 12 hours of logs for historical activity
  • Includes it in future scan cycles

Approving Actions

When Hiro proposes an action, you’ll see:
  • Action description — What will be done
  • Target — The affected user, resource, or system
  • Reason — Why this action is recommended
  • Evidence — Supporting findings from the investigation
  • Confidence score — How certain Hiro is about this action
All high-impact actions require explicit approval. Hiro will never suspend users, terminate instances, or revoke access without your consent.

Approval Workflow

  1. Review the action — Read the description, reason, and evidence
  2. Check confidence — Higher confidence means stronger evidence
  3. Decide:
    • Approve — Execute the action immediately
    • Reject — Skip this action, continue scanning
    • More Info — Ask Hiro for additional context

Session Persistence

Fight Mode sessions are saved automatically. You can:
  • Pause and resume — Close your browser and continue later
  • Review history — Access completed sessions from Fight Mode > History
  • Export findings — Download a full report of the session
Sessions include a complete audit trail of all findings, proposed actions, and executed remediations—useful for post-incident reviews.

Best Practices

Start with Specific Indicators

# Good
Investigate IP 185.220.101.1 and user john@company.com

# Less effective
Something seems wrong with our Okta logs

Provide Context

Investigate IP 185.220.101.1. This IP was flagged by our firewall
at 2pm today. The user john@company.com may be compromised.

Don’t Over-Scope

Focus Fight Mode on specific incidents. For broad security posture analysis, use the Chat interface instead.

Review Before Approving

Even with high confidence scores, take a moment to review the evidence before approving destructive actions.

Transitioning from Chat

You can escalate from a Chat investigation to Fight Mode:
This looks like an active attack. Start Fight Mode with the
indicators we've discussed.
Hiro will:
  1. Create a new Fight Mode session
  2. Import all discovered indicators
  3. Begin continuous scanning immediately

Next Steps

Remediation Actions

See all available actions and approval levels.

Detections

Learn how Hiro ingests and manages security detections.