BOD 26-04 Critical Vulnerability Triage

High severity TLP:CLEAR Research by Rapid7 Published 18 Nov 2026

Hypothesis

An attacker has exploited a publicly exposed, KEV-listed application (SharePoint, PAN-OS, or N-central) before remediation, establishing initial access or executing code with high privileges.

Brief

Why Now

Rapid7 recently highlighted the urgency of CISA Binding Operational Directive (BOD) 26-04 in "The Race Against the 3-Day Clock" (https://www.rapid7.com/blog/post/it-racing-the-three-day-clock-cisa-bod-26-04). This directive mandates that organizations patch Tier 1 vulnerabilities within 72 hours. Crucially, it also requires a concurrent forensic investigation to detect pre-patch compromise. This hunt provides the structure to meet that investigative requirement for vulnerabilities in SharePoint, PAN-OS, and N-central.

How the Hunt Flows

The hunt begins by querying the hb_vulnerability_finding surface. It filters for specific CVE IDs associated with the target applications. The results provide a list of vulnerable assets and their severity status. An analyst uses these identifiers to populate the scope for subsequent behavioral steps. This ensures the hunt focuses forensic resources on the most at-risk systems rather than scanning the entire estate indiscriminately. The second phase pivots to the hb_http_activity surface. It monitors sensitive URL paths, such as SharePoint layouts or management portals, for anomalous status codes. The query specifically isolates 403 Forbidden and 500 Internal Server Error responses that occur with low frequency. These errors often represent the noise of exploitation, where an attacker’s initial bypass attempts or malformed payloads fail before they achieve a successful breach. The hunt then examines the hb_process_activity surface to find high-fidelity evidence of exploitation. It looks for common shells—including cmd.exe, powershell.exe, and bash—running as child processes of vulnerable service parents. For example, a shell spawned by w3wp.exe on a SharePoint server or tomcat on an N-central appliance is a clear indicator of a command injection or remote code execution event. A specialized agent synthesizes the results from the scoping and behavioral phases. It correlates the vulnerability data with the observed activity to produce a compromise verdict for each host. This synthesis fulfills Phase 5 of the CISA protocol. The output distinguishes between benign service behavior, suspicious failed attempts, and confirmed malicious activity, allowing the security team to decide if they must initiate a full incident response.

Blind Spots

This hunt cannot detect sophisticated exploitation that occurs entirely within volatile memory. If a buffer overflow allows an attacker to inject and execute code inside the memory space of a running process without spawning a new shell, the process telemetry will not capture the event. Forensic RAM capture is required for that level of visibility. Additionally, the hunt does not inspect HTTP request bodies. Attacks that use valid-looking requests to bypass authentication via logic flaws may not trigger the 403 or 500 error patterns monitored here.

Steps

  1. Query, scoping

    Identify KEV-listed vulnerable assets

    Scope the estate to assets reporting target CVEs. While this surface uses device_uid, the agent will correlate this with hostnames from hb_devices to enable downstream triage on behavioral surfaces.

    sql, reads hb_vulnerability_finding
    SELECT device_uid, resource_uid, cve_uid, severity, status, first_seenFROM hb_vulnerability_findingWHERE instr(',' || '{{cve_ids}}' || ',', ',' || cve_uid || ',') > 0  AND status != 'suppressed'  AND resource_type IN ('device', 'virtual_machine')

    What a hit looks like. A list of device identifiers that are confirmed vulnerable. Silence indicates no known vulnerable assets in current scan results.

  2. Query, baseline

    Rare HTTP errors on vulnerable paths

    Find unusual internal server errors or forbidden responses on the vulnerable portals, which often correlate with bypass attempts or failed exploit payloads.

    sql, reads hb_http_activity
    SELECT device_hostname, url_path, status_code, COUNT(DISTINCT device_hostname) AS hosts, COUNT(*) AS requestsFROM hb_http_activityWHERE (status_code = 403 OR status_code = 500)  AND instr(',' || '{{vulnerable_paths}}' || ',', ',' || LOWER(url_path) || ',') > 0  AND (('{{scope_hosts}}' = '') OR (instr(',' || '{{scope_hosts}}' || ',', ',' || device_hostname || ',') > 0))  AND time >= datetime('now', '-{{lookback_days}} days')GROUP BY device_hostname, url_path, status_codeHAVING requests < 10ORDER BY requests ASC

    What a hit looks like. Anomalous 403 or 500 status codes on specific portal paths. Silence suggests no obvious failed exploit traffic matching these indicators.

  3. Query, detection candidate

    Shells spawned by vulnerable services

    Detect arbitrary code execution by identifying shells launched from the vulnerable service parents, including the Palo Alto and N-central daemons.

    sql, reads hb_process_activity
    SELECT device_hostname, process_name, process_cmd_line, parent_process_name, user_name, timeFROM hb_process_activityWHERE (('{{scope_hosts}}' = '') OR (instr(',' || '{{scope_hosts}}' || ',', ',' || device_hostname || ',') > 0))  AND instr(',' || '{{shell_names}}' || ',', ',' || LOWER(process_name) || ',') > 0  AND instr(',' || '{{vulnerable_parents}}' || ',', ',' || LOWER(parent_process_name) || ',') > 0  AND time >= datetime('now', '-{{lookback_days}} days')

    What a hit looks like. A shell process running with service-level or root privileges. This is a high-fidelity indicator of exploitation.

  4. Agent triage

    Synthesize triage evidence

    Analyze the vulnerability scoping and behavioral results to reach a compromise verdict per host, fulfilling Phase 5 of the CISA protocol.

  5. Decision

    Evaluate compromise verdict

    Route the hunt based on whether evidence of compromise was found during the triage window.

  6. Analyst task

    Remediation and escalation review

    Analyze the findings to determine if the vulnerability has been exploited and if full IR escalation is required.

  7. Analyst task

    Compliance close-out

    Document the forensic triage results to satisfy BOD 26-04 reporting requirements.

Coverage

3 of 4 stages covered

Stage and ATT&CK techniqueCoveredHow, or why not
Public Asset Exposure
T1190
Yes vulnerability-scoping
Vulnerability Exploitation
T1190
Yes rare-portal-errors
System-Level Execution
T1059
Yes service-shell-execution
Post-Exploitation Activity
T1021 T1083 T1074
Not visible Requires volatile memory analysis or network traffic pcaps which are not present in the listed surfaces.

Blind spots

  • Needs Forensic RAM capture (e.g. Velociraptor). A sophisticated buffer overflow (like CVE-2026-0300) can execute code entirely in memory; without volatile data capture, this hunt may return a false negative. It would answer: Was fileless code injected into memory without spawning a separate process?
  • Needs hb_http_activity with request_body or WAF logs. Surface telemetry sees only metadata; if the exploit doesn't trigger a 403/500 or a child process, it remains invisible. It would answer: Did the HTTP request contain a JWT bypass or exploit payload in the body?

Parameters and data

Parameters

ParameterTypeDefaultWhat it is
cve_idslist[string]CVE-2026-0300, CVE-2026-18556, CVE-2026-18577, CVE-2026-55040, CVE-2026-63520Target KEV-listed vulnerabilities from the CISA catalog.
lookback_daysnumber14Days of history to examine for exploitation evidence.
scope_hostslist[host]noneHostnames identified as vulnerable in the scoping step; leave empty to scan the entire estate.
shell_nameslist[string]cmd.exe, powershell.exe, sh, bash, zshStandard shell processes often used for arbitrary code execution.
vulnerable_parentslist[string]w3wp.exe, java, pan_auth, authd, tomcatProcesses associated with the vulnerable services (SharePoint, PAN-OS, N-central).
vulnerable_pathslist[path]/portal/, /_layouts/15/, /ncentral/Sensitive URL paths associated with the vulnerable service portals.

Data sources

SourceCategoryTelemetry
Endpoint telemetry (hb_ surfaces)endpointendpoint
Web server / proxy logssiemnetwork

Source

An open hunt.md file. It runs anywhere that reads the format.

hunt.md
---
analysis: A detection rule fires on a known shell; this hunt performs forensic triage
  by cross-referencing vulnerability scan status, internet exposure, and anomalous
  server-side error patterns to satisfy a specific regulatory mandate.
blind_spots:
- id: no-volatile-memory-access
  question: Was fileless code injected into memory without spawning a separate process?
  requires: Forensic RAM capture (e.g. Velociraptor)
  risk: A sophisticated buffer overflow (like CVE-2026-0300) can execute code entirely
    in memory; without volatile data capture, this hunt may return a false negative.
  stage: post-exploitation-triage
- id: http-body-inspection
  question: Did the HTTP request contain a JWT bypass or exploit payload in the body?
  requires: hb_http_activity with request_body or WAF logs
  risk: Surface telemetry sees only metadata; if the exploit doesn't trigger a 403/500
    or a child process, it remains invisible.
  stage: unauthenticated-exploit-bypass
coverage:
- stage: exposed-vulnerable-applications
  status: covered
  steps:
  - vulnerability-scoping
- stage: unauthenticated-exploit-bypass
  status: covered
  steps:
  - rare-portal-errors
- stage: elevated-execution
  status: covered
  steps:
  - service-shell-execution
- blind_spot: no-volatile-memory-access
  reason: Requires volatile memory analysis or network traffic pcaps which are not
    present in the listed surfaces.
  stage: post-exploitation-triage
  status: not_visible
guardrails:
  claims: no_unsupported
  evidence: citation_required
  missing_data: not_benign
  telemetry: untrusted
hunt:
  applicability: campaign-specific
  handoff: keep-as-periodic-hunt
  justification: CISA BOD 26-04 mandates that organizations must run a forensic investigation
    in parallel with patching for Tier 1 vulnerabilities. Failure to do so leaves
    an organization blind to intruders who gained access before the patch was applied.
  methodology: model-assisted
  trigger: intel-report
hypothesis: An attacker has exploited a publicly exposed, KEV-listed application (SharePoint,
  PAN-OS, or N-central) before remediation, establishing initial access or executing
  code with high privileges.
labels:
- hunt
- attack.t1190
- attack.t1059.003
- attack.t1059.004
- attack.t1021.001
- attack.t1074.001
- execution
- initial access
- lateral movement
name: BOD 26-04 Critical Vulnerability Triage
parameters:
  cve_ids:
    default:
    - CVE-2026-0300
    - CVE-2026-18556
    - CVE-2026-18577
    - CVE-2026-55040
    - CVE-2026-63520
    description: Target KEV-listed vulnerabilities from the CISA catalog.
    from:
      kind: article
      observed: '2026-11-04'
      ref: Rapid7 BOD 26-04
    type: list[string]
  lookback_days:
    default: '14'
    description: Days of history to examine for exploitation evidence.
    type: number
  scope_hosts:
    default: []
    description: Hostnames identified as vulnerable in the scoping step; leave empty
      to scan the entire estate.
    type: list[host]
  shell_names:
    default:
    - cmd.exe
    - powershell.exe
    - sh
    - bash
    - zsh
    description: Standard shell processes often used for arbitrary code execution.
    type: list[string]
  vulnerable_parents:
    default:
    - w3wp.exe
    - java
    - pan_auth
    - authd
    - tomcat
    description: Processes associated with the vulnerable services (SharePoint, PAN-OS,
      N-central).
    type: list[string]
  vulnerable_paths:
    default:
    - /portal/
    - /_layouts/15/
    - /ncentral/
    description: Sensitive URL paths associated with the vulnerable service portals.
    type: list[path]
provenance:
  authors:
  - name: Huntbase hunt generation
    org: huntbase.io
  generated:
    by: huntbase-hunt-generation
    from: https://www.rapid7.com/blog/post/it-racing-the-three-day-clock-cisa-bod-26-04
    gates:
    - dry-run
    - lint
    model: hb_google/gemini-3-flash-preview
rationale: Focus first on hosts confirmed as internet-facing in hb_exposed_assets.
  Use the device_uid from hb_vulnerability_finding to retrieve the Network Hostname
  from hb_devices to correctly populate scope_hosts for the behavioral steps.
references:
- name: "Rapid7 \u2014 The Race Against the 3-Day Clock: How Rapid7 Helps Meet BOD\
    \ 26-04 Compliance"
  url: https://www.rapid7.com/blog/post/it-racing-the-three-day-clock-cisa-bod-26-04
related:
- hunt: shadow-it-exposed-service-discovery
  reason: Correct identification of the external attack surface is required to accurately
    scope BOD 26-04 timelines.
  relation: precedes
scenario:
  stages:
  - name: Public Asset Exposure
    observables:
    - microsoft 365
    - palo alto
    - sharepoint
    - n-central
    - User-ID Authentication Portal
    - Captive Portal
    - CVE-2026-55040
    - CVE-2026-63520
    - CVE-2026-0300
    - CVE-2026-18556
    - CVE-2026-18577
    slug: exposed-vulnerable-applications
    tactic: initial-access
    techniques:
    - T1190
  - name: Vulnerability Exploitation
    observables:
    - JWT bypass
    - buffer overflow
    - specially crafted packets
    - authentication bypass
    - account takeover
    slug: unauthenticated-exploit-bypass
    tactic: initial-access
    techniques:
    - T1190
  - name: System-Level Execution
    observables:
    - SharePoint service account
    - root privileges
    - arbitrary code execution
    - unauthenticated RCE
    slug: elevated-execution
    tactic: execution
    techniques:
    - T1059
  - name: Post-Exploitation Activity
    observables:
    - lateral movement
    - data staging
    - volatile data capture
    - persistence establishment
    slug: post-exploitation-triage
    tactic: lateral-movement
    techniques:
    - T1021
    - T1083
    - T1074
  summary: Adversaries exploit publicly exposed, high-impact vulnerabilities in SharePoint,
    PAN-OS, and N-central to achieve unauthenticated remote code execution or administrative
    takeover. Under CISA BOD 26-04, these Tier 1 exploits require rapid forensic triage
    within 72 hours to detect post-exploitation behaviors such as lateral movement
    and data staging.
severity: high
targets:
  analyst:
    name: Tier-2 analyst
    role: analyst
  endpoint:
    category: endpoint
    name: Endpoint telemetry (hb_ surfaces)
    telemetry:
    - endpoint
  hunter:
    agent: true
    name: Hunt agent
  web:
    category: siem
    name: Web server / proxy logs
    telemetry:
    - network
tlp: clear
type: investigation
---


# BOD 26-04 Critical Vulnerability Triage

This hunt implements the mandatory forensic triage lifecycle established by CISA Binding Operational Directive 26-04. The directive requires that for Tier 1 vulnerabilities (actively exploited and providing total control), organizations must not only patch within 72 hours but also perform a concurrent forensic investigation to detect pre-patch compromise. This hunt scopes the estate to vulnerable assets, identifies anomalous web activity on those portals (specifically 403/500 errors indicating bypass attempts), and monitors for shells spawned by the vulnerable service daemons. An agent synthesizes the evidence to meet the triage reporting requirement.

## vulnerability-scoping
<!-- Identify KEV-listed vulnerable assets -->
Scope the estate to assets reporting target CVEs. While this surface uses device_uid, the agent will correlate this with hostnames from hb_devices to enable downstream triage on behavioral surfaces.

```sqlite target=endpoint role=scoping params=(cve_ids=cve_ids)
~~~yaml
expected: A list of device identifiers that are confirmed vulnerable. Silence indicates
  no known vulnerable assets in current scan results.
reads:
- device_uid
- resource_uid
- cve_uid
- severity
- status
- first_seen
silence: not_evidence_of_absence
source: hb_vulnerability_finding
verified: dry-run
verified_at: '2026-11-18'
~~~
SELECT device_uid, resource_uid, cve_uid, severity, status, first_seen FROM hb_vulnerability_finding WHERE instr(',' || '{{cve_ids}}' || ',', ',' || cve_uid || ',') > 0 AND status != 'suppressed' AND resource_type IN ('device', 'virtual_machine')
```

## rare-portal-errors
<!-- Rare HTTP errors on vulnerable paths -->
Find unusual internal server errors or forbidden responses on the vulnerable portals, which often correlate with bypass attempts or failed exploit payloads.

```sqlite target=web role=baseline params=(vulnerable_paths=vulnerable_paths, scope_hosts=scope_hosts, lookback_days=lookback_days)
~~~yaml
baseline:
  compare: new_this_window
  window: '{{lookback_days}}d'
expected: Anomalous 403 or 500 status codes on specific portal paths. Silence suggests
  no obvious failed exploit traffic matching these indicators.
prevalence:
  by: device_hostname
  key:
  - url_path
  - status_code
  rare_below: 10
reads:
- device_hostname
- url_path
- status_code
- time
silence: not_evidence_of_absence
source: hb_http_activity
verified: dry-run
verified_at: '2026-11-18'
~~~
SELECT device_hostname, url_path, status_code, COUNT(DISTINCT device_hostname) AS hosts, COUNT(*) AS requests FROM hb_http_activity WHERE (status_code = 403 OR status_code = 500) AND instr(',' || '{{vulnerable_paths}}' || ',', ',' || LOWER(url_path) || ',') > 0 AND (('{{scope_hosts}}' = '') OR (instr(',' || '{{scope_hosts}}' || ',', ',' || device_hostname || ',') > 0)) AND time >= datetime('now', '-{{lookback_days}} days') GROUP BY device_hostname, url_path, status_code HAVING requests < 10 ORDER BY requests ASC
```

## service-shell-execution
<!-- Shells spawned by vulnerable services -->
Detect arbitrary code execution by identifying shells launched from the vulnerable service parents, including the Palo Alto and N-central daemons.

```sqlite target=endpoint role=detection-candidate params=(scope_hosts=scope_hosts, shell_names=shell_names, vulnerable_parents=vulnerable_parents, lookback_days=lookback_days)
~~~yaml
expected: A shell process running with service-level or root privileges. This is a
  high-fidelity indicator of exploitation.
reads:
- device_hostname
- process_name
- process_cmd_line
- parent_process_name
- user_name
- time
silence: not_evidence_of_absence
source: hb_process_activity
verified: dry-run
verified_at: '2026-11-18'
~~~
SELECT device_hostname, process_name, process_cmd_line, parent_process_name, user_name, time FROM hb_process_activity WHERE (('{{scope_hosts}}' = '') OR (instr(',' || '{{scope_hosts}}' || ',', ',' || device_hostname || ',') > 0)) AND instr(',' || '{{shell_names}}' || ',', ',' || LOWER(process_name) || ',') > 0 AND instr(',' || '{{vulnerable_parents}}' || ',', ',' || LOWER(parent_process_name) || ',') > 0 AND time >= datetime('now', '-{{lookback_days}} days')
```

## forensic-triage-agent
<!-- Synthesize triage evidence -->
```agent target=hunter
cite: required
context:
- vulnerability-scoping
- rare-portal-errors
- service-shell-execution
max_iterations: 5
objective: Determine if any host identified as vulnerable (via device_uid correlation)
  shows evidence of exploitation in the HTTP or process logs. The verdict must distinguish
  between failed attempts (403/500 errors) and successful code execution (shells spawned
  by service parents).
success_criteria: A verdict of malicious | suspicious | benign per host, citing specific
  rows for any compromise evidence found.
tools:
- endpoint
- web
```

## triage-decision
<!-- Evaluate compromise verdict -->
if~: "the forensic-triage-agent verdict is malicious or suspicious for at least one host" (confidence: high, judge=hunter)
then: → remediation-review
indeterminate: → remediation-review
unavailable: → remediation-review (blind_spot: no-volatile-memory-access)
else: → compliance-close-out

## remediation-review
<!-- Remediation and escalation review -->
```manual target=analyst
Review the agent's cited rows. If exploitation is confirmed, activate the Incident Response plan immediately (Phase 6 of BOD 26-04). If only failed attempts were found, ensure the patching team has prioritized these specific assets for immediate remediation to close the exposure window.
```
→ compliance-close-out

## compliance-close-out
<!-- Compliance close-out -->
```manual target=analyst
Record the hosts examined, the search criteria used (403/500 errors and service-child processes), and the final forensic verdict. This documentation serves as the organization's proof of compliance with the 72-hour triage mandate.
```
→ end

Run this hunt against your own data.

In Huntbase, Scout runs every step across the tools you already connect, checks back through your history and returns a verdict with the evidence. The hunt.md file is open, so you can also take it anywhere that reads the format.

Drafted by machine (huntbase-hunt-generation, hb_google/gemini-3-flash-preview), checked by dry-run, lint, then reviewed by a person.

Content pack

Sign up to open

Pack contents open in Huntbase.