Critical TeamCity Flaw Could Let Attackers Run OS Commands Without Logging In

1. Executive summary

JetBrains has disclosed CVE-2026-63077, a CRITICAL vulnerability (CVSS 9.8, CWE-502 — Deserialization of Untrusted Data) affecting all TeamCity On-Premises versions. The flaw permits unauthenticated remote code execution via the agent polling protocol, enabling an attacker with HTTP(S) access to bypass authentication and execute arbitrary OS commands with the privileges of the TeamCity server process. Fixed versions 2025.11.7 and 2026.1.3 are available, alongside a security patch plugin for versions 2017.1+. There is no evidence of in-the-wild exploitation at time of publication. EMEA financial services running on-premise TeamCity as a CI/CD pipeline component face potential exposure of build configurations, stored credentials, and source code, as well as the risk of supply-chain compromise through build pipeline tampering.

2. Regulatory framing

3. Technical analysis & attack chain

Confirmed attack chain

1. Reconnaissance / access: The attacker requires HTTP(S) network access to the TeamCity server. Internet-facing instances are directly exposed; internal instances require prior network access.
2. Authentication bypass: The attacker targets the TeamCity agent polling protocol. This protocol is used for build-agent-to-server communication and is not subject to the same authentication checks as the web UI or REST API. The vulnerability (CWE-502 — deserialization of untrusted data) allows the attacker to send a crafted payload through this channel that bypasses authentication entirely.
3. Code execution: The deserialized payload executes arbitrary operating system commands with the privileges of the TeamCity server process. The severity is CVSS 9.8 with a network vector, no privileges required, and no user interaction.
4. Post-compromise impact: Depending on the OS-level privileges of the TeamCity service account, the attacker can: - Access and exfiltrate TeamCity data, build configurations, and stored credentials (including VCS tokens, SSH keys, and API keys stored in the CI system). - Modify server state, including build pipeline configurations — enabling supply-chain attacks via modified build artefacts or injected malicious build steps.

Additional vulnerabilities in the same advisory window: The BSI (WID-SEC-2026-1758, WID-SEC-2026-2504) and GitHub Security Advisories document multiple related TeamCity flaws addressed in the same release cycle, including arbitrary file access via Perforce VCS integration (pre-2026.1.2), RCE via Perforce connection settings (pre-2026.1), improper permission checks exposing build configuration parameters (pre-2026.1), and pipeline modification due to improper permission checks (pre-2026.1.2). These are distinct from CVE-2026-63077 but reinforce the urgency of upgrading to the latest fixed version rather than applying only the single-CVE patch plugin.

Confidence caveat: The primary technical detail (agent polling protocol as the attack vector, CWE-502 deserialization mechanism) is sourced from JetBrains via The Hacker News. The verified NVD data confirms CWE-502 and CVSS 9.8. No IOCs, exploit code, or threat actor attribution have been published. There is no evidence of exploitation in the wild per JetBrains' statement.

4. Mitigation & containment

P1 — Within 24 hours

• Inventory and assess exposure: Identify all TeamCity On-Premises instances in the estate. Determine whether any are internet-facing (HTTP/HTTPS reachable from outside the corporate network).
• Restrict network access: Immediately block external access to TeamCity servers at the network perimeter. If the agent polling port cannot be segregated from general web access, place the entire TeamCity server behind VPN or a zero-trust network access gateway. JetBrains explicitly warns that even exposing the login screen or REST API provides attackers with entry points.
• Apply the security patch plugin: For instances that cannot be immediately upgraded, install the JetBrains security patch plugin for versions 2017.1+. This addresses CVE-2026-63077 only. Download from JetBrains' official advisory page.

P2 — Within 72 hours

• Upgrade to a fixed version: Update all TeamCity On-Premises instances to version 2025.11.7 or 2026.1.3 (or later). This addresses CVE-2026-63077 and the additional related vulnerabilities (Perforce VCS file access, Perforce RCE, permission check bypasses, pipeline modification) documented in the GitHub advisories and BSI alerts.
• Review service account privileges: Audit the OS-level account running the TeamCity server process. If it runs with elevated privileges (e.g., root on Linux or Administrator on Windows), reduce to a least-privilege service account.
• Rotate stored credentials: If any TeamCity instance was internet-facing or accessible to broad internal networks prior to patching, rotate all credentials stored in TeamCity: VCS tokens, SSH keys, API keys, and any secrets used in build pipelines. Treat stored credentials as potentially compromised.

P3 — Within 7 days

• Harden the CI/CD attack surface: Implement network segmentation so that the agent polling protocol is only reachable from known build agent IP addresses. Restrict the TeamCity web UI to authenticated users on a management network.
• Verify build integrity: For instances that were exposed, review recent build configurations and pipeline definitions for unauthorised modifications. Compare build artefacts against known-good baselines.
• Update third-party risk records: Record the advisory and remediation status in the organisation's ICT third-party risk register for JetBrains products.

5. Indicators of compromise

No indicators of compromise available in the source material. JetBrains states there is no evidence of exploitation in the wild, and no IOCs, exploit artefacts, or attacker infrastructure have been published.

Behavioural indicators

6. Detection

Insufficient indicators to author detection rules. The sources do not contain exploit artefacts, file hashes, distinctive command-line strings, registry keys, or network signatures specific to this vulnerability. The behavioural indicators in §5 should be implemented as custom log-analysis rules in the organisation's SIEM using the described patterns.

CVE assessment

1 referenced CVE — 1 critical (CVSS ≥ 9.0)

7. Sources

• The Hacker News — Critical TeamCity Flaw Could Let Attackers Run OS Commands Without Logging In — https://thehackernews.com/2026/07/critical-teamcity-flaw-could-let.html — 2026-07-28
• GitHub Security Advisories — GHSA-j8x6-f4hm-gpqh: JetBrains TeamCity arbitrary file access via Perforce VCS — https://github.com/advisories/GHSA-j8x6-f4hm-gpqh
• GitHub Security Advisories — GHSA-pqrm-v5f6-j44f: JetBrains TeamCity improper permission checks exposing build configuration — https://github.com/advisories/GHSA-pqrm-v5f6-j44f
• GitHub Security Advisories — GHSA-hqxf-vmvp-qr3c: JetBrains TeamCity RCE via Perforce connection settings — https://github.com/advisories/GHSA-hqxf-vmvp-qr3c
• GitHub Security Advisories — GHSA-6wf4-qp9f-c3xh: JetBrains TeamCity pipeline modification via improper permission checks — https://github.com/advisories/GHSA-6wf4-qp9f-c3xh
• BSI Germany — WID-SEC-2026-1758: JetBrains TeamCity: Mehrere Schwachstellen — https://wid.cert-bund.de/portal/wid/securityadvisory?name=WID-SEC-2026-1758
• BSI Germany — WID-SEC-2026-2504: JetBrains TeamCity: Mehrere Schwachstellen ermöglichen Codeausführung — https://wid.cert-bund.de/portal/wid/securityadvisory?name=WID-SEC-2026-2504

8. Adverse Trace position

CVE-2026-63077 is a maximum-severity (CVSS 9.8 CRITICAL) unauthenticated remote code execution vulnerability in a widely used CI/CD platform. The attack vector — the agent polling protocol — does not require valid credentials, making internet-facing TeamCity instances immediately exploitable. For EMEA financial services, the risk extends beyond the TeamCity server itself: stored VCS credentials, SSH keys, and API tokens in the CI system provide lateral movement paths into source-code repositories, deployment infrastructure, and cloud environments. Build pipeline tampering could enable supply-chain compromise of production software. We assess the risk as HIGH for any client running TeamCity On-Premises, particularly instances that are internet-facing or on broadly accessible internal networks. There is no confirmed in-the-wild exploitation and no attributed threat actor at time of publication. Adverse Trace will monitor for exploit code publication, KEV addition by CISA, and any threat actor adoption. Clients should treat this as a P1 remediation and report to us any detection of anomalous agent polling activity or unauthorised build configuration changes on TeamCity hosts.

Read the original source →

Published via PulseTrace — Adverse Trace threat intelligence.