# Federal Agencies Get 72 Hours to Patch Exploited Flaws in Langflow, N-central, and Apache Tomcat


## The Threat


Three separate software vulnerabilities — spanning an AI development framework, an enterprise remote monitoring platform, and the world's most widely deployed Java web server — are being actively weaponized against real targets. CISA's addition of all three to its Known Exploited Vulnerabilities catalog, with a patch deadline of Friday, August 8, underscores that these aren't theoretical risks: defenders are already behind.


The most alarming of the three hits IBM's Langflow, a visual framework for building AI agent pipelines that has surged in popularity as organizations race to deploy agentic AI applications. CVE-2026-9198 carries a CVSS score of 9.8 and requires no credentials — attackers chain two API endpoints to bypass authentication entirely and drop arbitrary code on the server. Fully functional proof-of-concept exploits have been circulating publicly since late July, and this isn't Langflow's first rodeo: CISA issued a separate emergency alert two weeks ago for CVE-2026-0770, another critical Langflow RCE that grants root-level access. Two critical RCEs in the same framework within a fortnight is a pattern that should prompt any AI team to audit their deployment surface immediately.


N-able's N-central platform adds a particularly uncomfortable dimension: the original patch for CVE-2026-18576 was insufficient, and threat actors found the gap before most customers had even applied the first fix. N-able detected active exploitation and pushed an emergency hotfix on Sunday, August 3rd. For managed service providers running N-central — a platform that by design has elevated access across client environments — an unauthenticated admin account takeover is a breach multiplier. One compromised instance can cascade into dozens of client networks.


## Severity and Impact


| CVE | Product | CVSS Score | Severity | Attack Vector | Complexity | Auth Required | CWE |

|-----|---------|-----------|----------|---------------|-----------|---------------|-----|

| CVE-2026-9198 | IBM Langflow | 9.8 | Critical | Network | Low | None | CWE-306 (Missing Auth for Critical Function) |

| CVE-2026-18576 | N-able N-central | High | High | Network | Low | None | CWE-287 (Improper Authentication) |

| CVE-2026-34486 | Apache Tomcat | 7.5 | High | Network | Low | None | CWE-311 (Missing Encryption of Sensitive Data) |


Note: CVE-2026-34486 is an incomplete remediation of CVE-2026-29146 (CVSS 9.8), meaning organizations that patched the original flaw may still be exposed.


## Affected Products


IBM Langflow

  • All default Langflow deployments vulnerable to CVE-2026-9198
  • CVE-2026-0770 (separate critical RCE) also under active exploitation

    • N-able N-central

  • All versions prior to N-central 2026.3 affected by CVE-2026-18576
  • Original patch for the vulnerability proved insufficient; emergency hotfix released August 3, 2026

    • Apache Tomcat

  • Versions affected by CVE-2026-34486 (incomplete fix for CVE-2026-29146)
  • Consult Apache's official security advisories for specific version ranges

    • ## Mitigations


    Langflow (CVE-2026-9198 and CVE-2026-0770)

  • Apply the latest IBM/Langflow security updates immediately — PoC exploits are public and weaponized
  • Restrict Langflow API exposure to trusted internal networks; do not expose the API interface to the public internet
  • Audit Langflow deployment logs for anomalous API calls chaining authentication and execution endpoints
  • Consider temporarily taking public-facing Langflow instances offline until patching is complete

    • N-able N-central (CVE-2026-18576)

  • Install the emergency hotfix released August 3, 2026 — upgrading to N-central 2026.3 or later is required
  • Audit all N-central administrative accounts for unauthorized access or configuration changes
  • Given the MSP context, notify downstream clients of potential exposure and review access logs across managed endpoints
  • Enable network-level controls to restrict N-central management interfaces to known IP ranges

    • Apache Tomcat (CVE-2026-34486)

  • Apply the latest Apache Tomcat security release — verify the patch specifically addresses CVE-2026-34486, not only the original CVE-2026-29146
  • Hunt for indicators of compromise consistent with reverse shell deployment; Unit 42 identified a Chinese-speaking threat actor conducting manual exploitation against Tomcat servers as recently as July 30
  • Review server logs for unexpected outbound connections and anomalous process spawning from Tomcat processes

    • CISA deadline for federal agencies: All mitigations must be applied by end of day Friday, August 8, 2026.


    ## References


  • [CISA Known Exploited Vulnerabilities Catalog](https://www.cisa.gov/known-exploited-vulnerabilities-catalog)
  • [CISA Advisory: Langflow RCE CVE-2026-0770](https://www.cisa.gov/)
  • [N-able Security Advisory: CVE-2026-18576](https://www.n-able.com/)
  • [Palo Alto Networks Unit 42: Apache Tomcat CVE-2026-34486 Exploitation Report](https://unit42.paloaltonetworks.com/)
  • [Apache Tomcat Security Advisories](https://tomcat.apache.org/security.html)

    • ---


    ## HackWire Analysis


    What makes this trio of KEV additions particularly revealing isn't any one vulnerability — it's the pattern. All three share a common characteristic: the exploitation window opened because someone moved too slowly or incompletely.


    N-central's situation is the most troubling. N-able's first patch didn't hold. Attackers found the gap, found a new path, and CISA had to issue an emergency deadline for a vulnerability that theoretically should have been closed weeks ago. This is the "patch bypass" problem security teams rarely talk about openly: fixing a vulnerability in complex enterprise software is genuinely hard, and rushing a patch under pressure increases the odds of shipping an incomplete one. For MSPs running N-central, the risk isn't just their own environment — it's every client network they manage. A single compromised N-central console can pivot laterally into hundreds of organizations. Ransomware operators know this, which is why RMM platforms have been high-priority targets since at least 2022.


    Langflow's back-to-back critical CVEs tell a different story: AI infrastructure is being built fast and secured slowly. The visual, low-code nature of Langflow makes it approachable for teams who aren't security-native, and those teams are often the ones deploying it on internet-exposed infrastructure without hardening defaults. With public PoCs already available and a second critical flaw still fresh, any Langflow instance accessible from outside a private network should be treated as compromised until proven otherwise.


    The Apache Tomcat exploitation by a Chinese-speaking threat actor installing reverse shells is a quieter signal worth watching. Manual, targeted campaigns against Java web infrastructure suggest reconnaissance and persistence — not opportunistic scanning. Organizations running Tomcat in government, defense, or critical infrastructure contexts should treat this as intelligence, not just a patch notification.


    — HackWire Editorial


    ## Related Coverage


  • Read more in our [Vulnerabilities](https://www.hackwire.news/category/vulnerabilities) coverage
  • Cross-reference with [Breaches](https://www.hackwire.news/category/breaches) and [Malware](https://www.hackwire.news/category/malware)
  • Stay current via the [HackWire homepage](https://www.hackwire.news/)