agentleFS
jaskaranhundaljaskaranhundal/usap-skills/response/CLAUDE.md

usap-skills / response

CLAUDE.md4 starsChanged 2 months agoApache-2.0View on GitHub

This file is the authoritative domain guide for Claude agents operating within the response/ skill domain of the USAP skills repository. It covers the purpose of the domain, the full skills catalog, Python tool references, incident response lifecycle procedures, severity matrices, regulatory deadline obligations, best practices, and cross-skill workflow sequences. The response domain provides skills for the full incident lifecycle from the moment a security event is detected through triage, active containment, digital forensics, post-incident review, and board-level communication. It…

The file

We split this CLAUDE.md into the rules it gives an agent

6 rules. About 0 are common, 3 are distinctive, and 3 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈3

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 3

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    All tools accept --help for parameter documentation.

    line 59
  2. Distinctive:

    Skills involved: All domain skills may be activated in parallel at this step depending on SEV level.

    line 85
  3. Very specific:

    If the incident involves a CVE with no available vendor patch, zero-day-response computes exposure scores for affected assets and …

    line 103
  4. Distinctive:

    Operational rule: If the incident scope is unclear at declaration time, assume the most restrictive applicable deadline until scope is …

    line 156
  5. Very specific:

    Containment approval is non-negotiable: No containment action that modifies system state (network isolation, credential revocation, …

    line 166
  6. Distinctive:

    Zero-day compensating controls are temporary by definition: Every compensating control deployed in the absence of a vendor patch must carry …

    line 174

≈ “Common” is an estimate: at least 1 in 5 of the 27,641 CLAUDE.md files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
CLAUDE.md

Response Domain — CLAUDE.md

This file is the authoritative domain guide for Claude agents operating within the response/ skill domain of the USAP skills repository. It covers the purpose of the domain, the full skills catalog, Python tool references, incident response lifecycle procedures, severity matrices, regulatory deadline obligations, best practices, and cross-skill workflow sequences.


Purpose

The response domain provides skills for the full incident lifecycle from the moment a security event is detected through triage, active containment, digital forensics, post-incident review, and board-level communication. It implements the Incident Command System (ICS) model adapted for cybersecurity operations, aligns with NIST SP 800-61 Rev 2, and integrates MITRE ATT&CK coverage across all response phases.

Domain responsibilities:

  • Incident command: Severity declaration (SEV1–SEV4), response track assignment, regulatory clock activation, and decision authority coordination across all response agents.
  • Triage: Universal first-triage classifying events into 14 incident types, severity scoring, false positive identification, and escalation routing.
  • Containment: Blast-radius-aware containment strategy selection across 10 threat types with production impact assessment and human approval enforcement.
  • Forensics: Legally defensible evidence collection, chain-of-custody compliance, DFRWS six-phase framework execution, and dwell time estimation.
  • Zero-day response: Exposure scoring, compensating control selection (5 options), vendor timeline tracking, and emergency change management when no patch is available.
  • Zero-day governance: Board and executive communication coordination, regulatory deadline tracking, and cross-organizational escalation matrix management.

Agents in this domain interact with the detection/ domain (upstream signal sources), the governance/ domain (regulatory notification and audit requirements), and the risk-compliance/ domain (findings lifecycle and vulnerability tracking).


Skills Catalog

SkillSlugPrimary ToolMITRE Coverage
incident-commanderresponse/incident-commanderincident-commander_tool.pyAll phases (command)
incident-classificationresponse/incident-classificationincident-classification_tool.pyTA0001–TA0010 (triage)
containment-advisorresponse/containment-advisorcontainment-advisor_tool.pyTA0005, TA0008 (containment)
forensicsresponse/forensicsforensics_tool.pyTA0009, TA0010 (collection)
zero-day-responseresponse/zero-day-responsezero-day-response_tool.pyT1190, T1203 (vuln exploitation)
zero-day-response-governanceresponse/zero-day-response-governancezero-day-response-governance_tool.pyGovernance layer

Skill Level and Plane Summary

SkillLevelPlaneApproval RequiredMutating Intents
incident-commanderL3workCISO / security_directornetwork_change, credential_operation, device_config_change, remediation_action
incident-classificationL3workNone (read_only)None
containment-advisorL3worksoc_lead, cisoremediation_action, network_change, credential_operation
forensicsL3workincident_commander (for evidence actions)remediation_action
zero-day-responseL3workCISO (for isolation); human approval for all controlsdevice_config_change, external_communication
zero-day-response-governanceL2controlCISO + Legal (for external notifications)external_communication

Python Tools Reference

Tool FileSkillInvocationOutput FormatPrimary Use
incident-commander_tool.pyincident-commanderpython scripts/incident-commander_tool.py --output jsonJSONSEV declaration, response track activation, regulatory clock
incident-classification_tool.pyincident-classificationpython scripts/incident-classification_tool.py --output jsonJSONEvent triage, type classification, false positive detection
containment-advisor_tool.pycontainment-advisorpython scripts/containment-advisor_tool.py --output jsonJSONContainment strategy, blast radius, production impact
forensics_tool.pyforensicspython scripts/forensics_tool.py --output jsonJSONEvidence collection, timeline reconstruction, chain of custody
zero-day-response_tool.pyzero-day-responsepython scripts/zero-day-response_tool.py --output jsonJSONExposure scoring, compensating controls, patch tracking
zero-day-response-governance_tool.pyzero-day-response-governancepython scripts/zero-day-response-governance_tool.py --output jsonJSONBoard communication, regulatory deadlines, escalation matrix

All tools accept --help for parameter documentation. All tools emit structured JSON compatible with the USAP evidence chain schema.


Incident Response Lifecycle

The response domain implements a seven-step procedure aligned with NIST SP 800-61 and the ICS model. Each step maps to one or more skills.

Step 1: Detection and Initial Notification

An event arrives from the detection/ domain (SIEM alert, EDR detection, threat intelligence feed, user report) or from an automated pipeline trigger. The incident-classification skill performs initial triage.

Skill invoked: incident-classification Output: incident_type, severity_assessment, false_positive_flag, escalation_recommendation

Step 2: Severity Declaration

Based on the classification output, incident-commander declares a severity level (SEV1–SEV4) and opens the incident command structure. The regulatory notification clock starts at this step.

Skill invoked: incident-commander Output: incident_severity, response_tracks, regulatory_notification_required, notification_deadline_utc

Step 3: Response Track Activation

The incident commander assigns parallel response tracks: containment, investigation, notification, and recovery. Each track is assigned to a USAP agent or human role.

Skills involved: All domain skills may be activated in parallel at this step depending on SEV level.

Step 4: Containment

containment-advisor assesses blast radius and recommends the most targeted containment action for the confirmed threat type. All containment actions that modify system state are mutating and require human approval before execution.

Skill invoked: containment-advisor Output: recommended_strategy, blast_radius, production_impact, urgency, mutating_category, approver_roles

Step 5: Forensic Collection and Evidence Preservation

Concurrent with containment, forensics initiates evidence collection following the DFRWS six-phase framework. Volatile evidence (memory, network connections, running processes) is captured first. All evidence items are hashed (SHA-256) and logged in the chain of custody.

Skill invoked: forensics Output: timeline, iocs_identified, evidence_preservation_actions, dwell_time_estimate, legal_hold_required

Step 6: Zero-Day Handling (Conditional)

If the incident involves a CVE with no available vendor patch, zero-day-response computes exposure scores for affected assets and recommends compensating controls from a five-option control set. zero-day-response-governance manages executive communication and regulatory tracking.

Skills invoked: zero-day-response, zero-day-response-governance Output: Exposure scoring table, compensating control plan, vendor timeline record, executive communication matrix

Step 7: Post-Incident Review

Following eradication and recovery, the incident record is finalized. The forensics output provides the technical timeline and evidence inventory. incident-commander closes the incident and triggers handoff to risk-compliance/ for findings tracking and lessons-learned integration.

Skills involved: forensics (final report), incident-commander (incident closure)


SEV Level Matrix

LevelNameCriteriaResponse SLASkills InvokedEscalation Path
SEV1CriticalConfirmed ransomware or destructive malware in production; active exfiltration of PII/PHI/financial data (>10K records); full network compromise or domain controller breach; defense evasion detected (CloudTrail disabled, SIEM wiped); supply chain compromise (build system breach)15 min bridge call; war room immediateincident-classification, incident-commander, containment-advisor, forensics (parallel)SOC Lead → CISO → CEO → Board Chair (if data at risk)
SEV2HighConfirmed unauthorized access to sensitive systems; credential compromise with elevated privileges; lateral movement detected and confirmed; ransomware indicators without confirmed execution1 hour bridge; async coordination within 30 minincident-classification, incident-commander, containment-advisor, forensicsSOC Lead → CISO
SEV3MediumSuspected unauthorized access (unconfirmed); malware detected and contained (no spread evidence); single account compromise (no privilege escalation)4 hours; async coordinationincident-classification, incident-commander, containment-advisorSOC Lead → Security Manager
SEV4LowSecurity alert with no confirmed impact; informational indicator; policy violation with no data risk24 hours; ticket-basedincident-classificationL3 Analyst queue

SEV Escalation Triggers

An incident must be immediately re-declared at a higher SEV level when any of the following indicators are observed:

IndicatorNew Severity
Ransomware note foundEscalate to SEV1
Active exfiltration confirmedEscalate to SEV1
CloudTrail or SIEM disabledEscalate to SEV1
Domain controller access confirmedEscalate to SEV1
Second system compromisedEscalate to SEV1
Exfiltration volume exceeds 1 GBEscalate to SEV2 minimum
C-suite account accessedEscalate to SEV2 minimum

Regulatory Notification Deadlines

The incident-commander and zero-day-response-governance skills track regulatory deadlines from the moment an incident is declared. The notification clock starts at declaration, not at investigation completion.

FrameworkIncident TypeDeadlineSkills Used
GDPR (EU 2016/679)Personal data breach72 hours after discoveryincident-commander, zero-day-response-governance
PCI-DSS v4.0Cardholder data breach24 hours after confirmation to acquirerincident-commander
HIPAA (45 CFR Part 164)PHI breach (>500 individuals)60 days after discovery; media notice if >500 in stateincident-commander, zero-day-response-governance
NY DFS 23 NYCRR 500Cybersecurity event72 hours to DFS; annual certificationincident-commander, zero-day-response-governance
SEC Cybersecurity Rule (17 CFR 229.106)Material cybersecurity incident4 business days after materiality determinationzero-day-response-governance
CCPA / CPRABreach of sensitive personal informationWithout unreasonable delay; notify AG if >500 CA residentsincident-commander, zero-day-response-governance
NIS2 (EU 2022/2555)Significant incident (operator of essential services)24-hour early warning; 72-hour notificationzero-day-response-governance
SOX (for material incidents)Financial system compromiseImmediate disclosure of material weaknesszero-day-response-governance

Operational rule: If the incident scope is unclear at declaration time, assume the most restrictive applicable deadline until scope is confirmed. Document the assumption in the incident record.


Domain Best Practices

  1. Decisiveness over perfection under time pressure. During a SEV1 incident, a good decision made at T+15 minutes outweighs a perfect decision made at T+45 minutes. All decisions must be logged in the evidence chain, including decisions made under uncertainty.
  2. Volatile evidence first. Memory, running processes, and active network connections are lost on system shutdown. The forensics skill must be invoked in parallel with containment — never sequentially after containment completes.
  3. Containment approval is non-negotiable. No containment action that modifies system state (network isolation, credential revocation, firewall rule change) may be executed without explicit human approval. The containment-advisor recommends; humans authorize; the tool-execution-broker executes.
  4. The notification clock starts at declaration. Regulatory deadlines run from the moment the incident is declared, not from investigation completion. When in doubt about regulatory scope, start the clock and confirm scope within the first response window.
  5. Chain of custody is established at evidence collection, not at report time. Every evidence item must be SHA-256 hashed at the moment of acquisition. Timestamps must be UTC with timezone offset. Tool provenance (FTK Imager, Volatility, AWS CloudTrail export) must be recorded per item.
  6. False positive verification before escalation. The incident-classification skill applies five false positive filters before escalating: known-safe automation, test environment activity, expected batch jobs, whitelisted identities, and scanner activity. Skipping these filters inflates SEV1 declaration rates and degrades SOC credibility.
  7. Zero-day compensating controls are temporary by definition. Every compensating control deployed in the absence of a vendor patch must carry a documented expiry trigger (patch release date or a specified quarterly review date). Controls that outlive their justification become permanent attack surface.
  8. Regulatory communication is privileged until legal review. All external communications related to a confirmed breach — to regulators, customers, or media — must be reviewed by Legal and CISO before transmission. zero-day-response-governance enforces this gate and must not be bypassed to meet a notification deadline; prepare draft notifications in advance and hold them pending approval.

Cross-Skill Workflow: SEV1 Response Track

The following sequence documents the canonical skill invocation order for a SEV1 incident. Steps 3 and 4 run in parallel.

T+0   Detection event arrives from detection/ domain
        |
        v
T+5   [incident-classification]
        Classify event type, score severity, check false positives
        Output: incident_type=ransomware, severity=critical, false_positive=false
        |
        v
T+10  [incident-commander]
        Declare SEV1, open war room, start regulatory clock
        Assign response tracks: containment / investigation / notification / recovery
        Output: incident_severity=sev1, regulatory_deadline=T+72h (GDPR), response_tracks=[...]
        |
        |-----------------------------------------------|
        v                                               v
T+15  [containment-advisor]                     [forensics]
        Identify blast radius                     Acquire volatile evidence (memory, netstat)
        Recommend isolation strategy              Start chain of custody log
        Requires approval: soc_lead + ciso        Hash all evidence at acquisition time
        Output: containment_plan, urgency=immediate    Output: timeline (partial), iocs_identified
        |                                               |
        v                                               v
T+30  [Human approval gate]                     [forensics — continued]
        Operator reviews containment plan         Disk image, CloudTrail, log export
        Approves or modifies                      Reconstruct attacker timeline
        |
        v
T+45  [tool-execution-broker]
        Executes approved containment actions
        |
        v
T+60  [incident-commander]
        Assess containment effectiveness
        Update regulatory notification status
        Brief Legal if PCI/GDPR scope confirmed
        |
        v
     [If zero-day involved]
        |
        v
     [zero-day-response]
        Score exposure for unpatched assets
        Select compensating control option (WAF / network block / feature disable / isolation / detection sensitivity)
        Track vendor patch timeline
        |
        v
     [zero-day-response-governance]
        Prepare board communication
        File regulatory notification (if required)
        Open emergency change record in ITSM
        |
        v
T+4h  Post-containment assessment
        [forensics] final evidence package, dwell time estimate
        [incident-commander] incident status update, next SLA checkpoint
        |
        v
      Eradication and recovery
        [incident-commander] close incident
        Handoff to risk-compliance/ for findings tracking

Related Domains

  • detection/: Upstream signal source. Provides SIEM alerts, EDR detections, and telemetry events that trigger incident-classification. The telemetry-signal-quality skill in detection/ assesses the fidelity of incoming signals before classification.
  • governance/: Downstream consumer of incident records. The compliance-mapping and internal-audit-assurance skills in governance/ receive incident summaries for regulatory reporting, audit evidence packages, and policy violation tracking.
  • risk-compliance/: Receives post-incident findings for vulnerability lifecycle management, lessons-learned integration, and recurrence prevention tracking.
  • agents/security/cs-incident-responder.md: The orchestrator agent that manages the full SEV1–SEV4 lifecycle by coordinating all skills in this domain.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.