CVE-2026-55428: Coder Tailnet IP Spoofing via AllowedIPs Validation Bypass

Technical summary

The vulnerability stems from inconsistent input validation in Coder's tailnet coordinator. While the coordinator correctly validates that an agent's `Addresses` field corresponds to the agent's authenticated UUID, no equivalent authorization check exists for the `AllowedIPs` parameter. An authenticated agent can supply arbitrary IP prefixes in `AllowedIPs`, which the coordinator forwards directly to tunnel peers without validation. These peers then install the attacker-controlled prefixes into their WireGuard interface configuration, creating unauthorized routing paths. The fix implements UUID-based validation for `AllowedIPs` identical to the existing `Addresses` validation, ensuring agents can only advertise IP ranges they are authorized to claim.

Business impact

In organizations using Coder for remote development infrastructure, this vulnerability enables lateral movement and traffic interception within the development network. An authenticated insider or compromised agent credential could redirect traffic destined for legitimate development environments, potentially exfiltrating code, stealing credentials, or executing man-in-the-middle attacks. The cross-scope aspect (CVSS S:C) indicates blast radius across logical network boundaries; in shared-tenant or multi-team Coder deployments, the risk extends beyond a single project or environment. Development teams may inadvertently route sensitive code or build artifacts through a malicious peer.

Affected systems

Coder versions prior to 2.29.7, 2.32.7, 2.33.8, and 2.34.2 are affected. Organizations should verify their deployed Coder version against these patched releases. The vulnerability affects any agent registered with a vulnerable coordinator, particularly in multi-agent deployments where agents communicate across team or organizational boundaries via a shared tailnet.

Exploitability

Exploitation requires valid Coder agent credentials (PR:L in CVSS). The attack is not trivial—an attacker must craft a malicious agent or modify an existing agent's configuration to advertise unauthorized `AllowedIPs`—hence the AC:H (high attack complexity) rating. However, once exploited, the impact is significant: the attacker gains visibility into peer traffic and can perform selective routing attacks without triggering immediate alerts. The barrier to exploitation is moderate; it is not a zero-click or unauthenticated vector, but any user with agent provisioning permissions or access to agent configuration could attempt this attack.

Remediation

Upgrade Coder to version 2.29.7, 2.32.7, 2.33.8, or 2.34.2 or later, depending on your current version line. Apply the patch during a maintenance window to avoid disrupting active development environments. For organizations unable to patch immediately, implement coordinator log monitoring to detect agents advertising unexpected `AllowedIPs` prefixes. Cross-reference advertised prefixes against your known network topology and approved agent configurations; any deviation warrants investigation.

Patch guidance

Verify your current Coder version using the CLI or administrative UI. Consult your Coder instance's release notes to determine which version line you are on (2.29.x, 2.32.x, 2.33.x, or 2.34.x), then upgrade to the corresponding patched version listed above. Test the patch in a staging environment if possible, particularly if you have custom networking or WireGuard peer configurations. After upgrading, restart or reconcile affected agents to apply the new validation rules.

Detection guidance

Enable verbose logging on the Coder coordinator and search for `AllowedIPs` entries that do not match the advertising agent's UUID-derived address ranges. Collect netflow or DNS logs from your development network to identify unexpected traffic routing patterns or tunneled flows to unfamiliar IP ranges. If you see agents advertising multiple disjoint `AllowedIPs` blocks, especially those corresponding to other teams' or organizations' subnets, escalate for manual review. WireGuard peer configuration audit tools can reveal mismatches between expected and installed peer settings.

Why prioritize this

This vulnerability merits rapid patching due to the HIGH CVSS score (8.2), high confidentiality and integrity impact, cross-scope risk, and the sensitivity of development environments. Development infrastructure often holds pre-production code, credentials, API keys, and other high-value secrets. A successful attack could compromise intellectual property or enable downstream attacks on production systems. Although PR:L limits the attack surface, the consequence of a breach is severe. Treat this as a priority 1 or 2 issue in your patch cycle.

Risk score, explained

CVSS 8.2 reflects high impact (C:H, I:H) limited by authentication requirement (PR:L) and attack complexity (AC:H). The cross-scope aspect (S:C) elevates risk in shared-tenant scenarios. No availability impact (A:N) prevents a perfect 9.0+, but the combination of confidentiality (traffic interception) and integrity (routing manipulation) in a network-sensitive context justifies the HIGH severity. For development-focused organizations where code and secrets are primary assets, the real-world risk may exceed the base CVSS if an attacker can persistently intercept or modify agent communications.

Frequently asked questions