/

CVE-2026-55436

Severity:

HIGH RISK

(7.4)

TLS Cert Validation Flaw (CWE-295) in Coder AI Proxy

CVE-2026-55436

Analyzed

Vulnerability type

Category:

Improper Certificate Validation

CWE:

CWE-295

Impact:

Man-in-the-middle traffic interception and tampering

Severity:

HIGH RISK

(7.4)

CVSS version:

3.1

Catch this CVE while code is being written

Detect vulnerable code directly in pull requests, so CVEs are fixed before merge.

Summary

Coder's AI Bridge Proxy used an HTTPS client configuration that disabled TLS certificate verification when no upstream proxy was set.

In this default state, an on-path attacker could impersonate the Coder server, intercept traffic, and potentially tamper with AI-related requests and responses.

Affected context

This vulnerability appears in Coder deployments using the AI Bridge Proxy (aibridgeproxyd) between versions 2.30.0 and prior to 2.32.7, 2.33.8, and 2.34.2.

It is relevant when the proxy connects to the Coder access URL over a network path where an attacker could gain an on-path position. Environments where the AI Bridge Proxy and Coder server are co-located on loopback, or protected with strong network segmentation or mutual TLS, are typically less exposed, but certificate validation still technically remains disabled in the vulnerable configuration.

Exploit preconditions

  • Attack surface: Outbound HTTPS connection from AI Bridge Proxy to the Coder access URL.
  • Required access: Ability to act as an on-path (man-in-the-middle) entity on the network between the AI Bridge Proxy and the Coder server.
  • Trigger point: AI Bridge Proxy initiating HTTPS requests using the default goproxy transport with InsecureSkipVerify: true and no upstream proxy configured.
  • Why exploitation is possible: TLS certificate verification is explicitly disabled in the default transport, allowing a malicious TLS endpoint with any certificate to be treated as the legitimate Coder server.

Why this happens

The bug stems from an HTTPS client configuration that intentionally disables certificate verification (InsecureSkipVerify: true) in the default transport for the goproxy server used by the AI Bridge Proxy.

This maps to CWE-295: Improper Certificate Validation, where the TLS stack fails to validate server certificates against a trusted root and hostname, breaking the fundamental assumption that TLS endpoints are authenticated. As a result, the proxy trusts any certificate presented by an attacker, enabling transparent interception or modification of traffic.

Potential impact

  • An on-path attacker may intercept and read AI-related requests and responses sent between the AI Bridge Proxy and the Coder server.
  • Attackers could tamper with model prompts, responses, or configuration data flowing through the proxy.
  • Credentials, tokens, or other sensitive metadata passed via the proxy can be exposed or misused.
  • Compromised responses could mislead developers or automate unsafe actions within remote development environments.

General mitigation guidance

Effective mitigation focuses on restoring strong TLS guarantees and constraining the network trust boundary between the AI Bridge Proxy and the Coder backend.

Common patterns include enforcing certificate validation for all HTTPS clients by default, using only transports that honor system or pinned root CAs, and avoiding configurations that explicitly disable verification.

Additionally, organizations can reduce exposure by isolating the proxy-to-server path (e.g., loopback, private network segments, mutual TLS, or service mesh with strict identity policies), and by treating any component that disables certificate checks as a high-risk configuration requiring immediate hardening.

Ensuring that service endpoints always present trusted certificates and that clients fail closed on validation errors helps maintain secure-by-default behavior.

How teams detect issues like this

Detection often relies on configuration and code analysis rather than runtime signatures.

Static analysis and code review can flag uses of TLS options like InsecureSkipVerify or equivalent flags that disable certificate validation. Reviewers can also look for custom transports that bypass system trust stores or hostname verification.

Configuration audits should verify that all service clients use validated HTTPS endpoints and that no component allows arbitrary certificates by policy.

Automated security tooling integrated into CI/CD pipelines can enforce rules disallowing insecure TLS settings and alert when such patterns are introduced or modified.

Sources

SourceLink
github.comgithub.com/coder/coder/pull/26131
github.comgithub.com/coder/coder/releases/tag/v2.32.7
github.comgithub.com/coder/coder/releases/tag/v2.33.8
github.comgithub.com/coder/coder/releases/tag/v2.34.2
github.comgithub.com/coder/coder/security/advisories/GHSA-84rm-42xw-mx52

Frequently asked

questions

It is exploitable only by an attacker who can position themselves on the network path between the AI Bridge Proxy and the Coder server. External remote attackers without this on-path capability generally cannot exploit it directly.

An attacker needs sufficient network control to act as a man-in-the-middle, such as compromised infrastructure, rogue Wi-Fi, or control over routing between the proxy and the server. Application-level credentials alone are not enough to exploit the TLS validation flaw.

The safest approach is to ensure all HTTPS clients perform strict certificate validation by default, using transports that honor system root CAs (or pinned roots) and hostname checks, and to remove any configuration that disables TLS verification.

Organizations can establish coding standards that forbid disabling certificate validation, enforce security-focused code review for network-related changes, and use automated static analysis to detect insecure TLS options. Adopting secure-by-default libraries and centralizing TLS configuration can further reduce the chance of similar vulnerabilities.

Find out what's already
exploitable in your codebase.