CVE-2026-63004: Unleash Feature Management SSRF Vulnerability

CVE-2026-63004 Overview

CVE-2026-63004 is a Server-Side Request Forgery (SSRF) vulnerability in Unleash, an open-source feature management platform. The flaw exists in the addon and integration subsystem, where operator-controlled parameters.url values are passed from src/lib/addons/webhook.ts and the Slack, Microsoft Teams, Datadog, and New Relic integrations directly into Addon.fetchRetry in src/lib/addons/addon.ts. The code does not restrict loopback, link-local, private, or cloud metadata addresses. An authenticated actor holding the root CREATE_ADDON or UPDATE_ADDON permission can force the server to issue requests inside the network boundary. This issue is fixed in versions 7.5.2, 7.6.5, and 8.0.2.

Critical Impact
Authenticated attackers with addon privileges can probe internal networks, exfiltrate authentication headers, and deliver feature-event data to internal services through SSRF [CWE-918].

Affected Products

Unleash versions prior to 7.5.2
Unleash versions prior to 7.6.5
Unleash versions prior to 8.0.2

Discovery Timeline

2026-08-21 - CVE-2026-63004 published to NVD
2026-08-21 - Last updated in NVD database

Technical Details for CVE-2026-63004

Vulnerability Analysis

The vulnerability resides in how Unleash processes URL parameters supplied to its integration and addon subsystem. The parameters.url field, defined by an operator with CREATE_ADDON or UPDATE_ADDON permissions, is forwarded to Addon.fetchRetry without allowlist enforcement or destination filtering.

Because the platform does not reject loopback (127.0.0.1), link-local (169.254.0.0/16), RFC1918 private ranges, or cloud metadata endpoints such as 169.254.169.254, an attacker can direct outbound requests at internal-only services. The integration event status returned by Unleash acts as a blind probing oracle, allowing the attacker to differentiate open, closed, and filtered internal ports.

A secondary risk is credential exposure. The Slack, Microsoft Teams, Datadog, and New Relic integrations transmit Authorization, customHeaders, or DD-API-KEY values along with the outbound request. Redirecting the URL to an attacker-controlled host exfiltrates these secrets. The full feature-event JSON body is also delivered to whichever endpoint the URL points at.

Root Cause

The root cause is missing destination validation in the addon HTTP client. The fetchRetry helper accepts any URL and does not enforce an outbound allowlist or deny-by-default policy for private networks, tracked as [CWE-918].

Attack Vector

Exploitation requires an authenticated account holding root CREATE_ADDON or UPDATE_ADDON permission. The attacker creates or updates a webhook, Slack, Teams, Datadog, or New Relic integration with a url targeting an internal service or cloud metadata endpoint. Trigger events cause the Unleash server to send authenticated requests to the specified destination.

typescript

// Patch excerpt: src/lib/addons/new-relic.ts
// Sensitive URL removed from event details to avoid leaking
// the operator-controlled destination in integration logs.
            stateDetails: stateDetails.join('\n'),
            event: serializeDates(event),
            details: {
-                url,
                body,
            },
        });
// Source: https://github.com/Unleash/unleash/commit/2100db76af3473f13e6fb40096cf17a9c2b741a1

Detection Methods for CVE-2026-63004

Indicators of Compromise

Addon or integration configurations containing url values pointing to 127.0.0.1, localhost, RFC1918 ranges, or 169.254.169.254.
Unusual outbound requests originating from the Unleash server to internal hosts or cloud metadata endpoints.
Audit log entries showing CREATE_ADDON or UPDATE_ADDON actions from unexpected accounts followed by rapid trigger events.

Detection Strategies

Review Unleash audit logs for addon create and update events referencing non-public URLs.
Inspect egress traffic from Unleash hosts using network flow logs and DNS query records for internal destinations.
Correlate integration event failure statuses with attempts to reach non-routable or metadata endpoints, indicating blind SSRF probing.

Monitoring Recommendations

Alert on any Unleash outbound connection to link-local, loopback, or private address space.
Monitor cloud provider metadata endpoint access from application workloads hosting Unleash.
Track privilege assignments granting CREATE_ADDON or UPDATE_ADDON to detect scope creep.

How to Mitigate CVE-2026-63004

Immediate Actions Required

Upgrade Unleash to 7.5.2, 7.6.5, or 8.0.2 as appropriate for your release branch.
Audit existing addon and integration configurations for URLs targeting internal ranges and remove them.
Restrict which accounts hold CREATE_ADDON and UPDATE_ADDON permissions to the minimum operational set.

Patch Information

The vendor released fixes in GitHub Release v7.5.2, GitHub Release v7.6.5, and GitHub Release v8.0.2. Technical details are documented in the GitHub Security Advisory GHSA-5vf6-jrqr-78fj and the patch commits 2100db7, d45f99d, and d862562.

Workarounds

Deploy Unleash behind an egress proxy that denies access to loopback, link-local, private, and cloud metadata address ranges.
Disable unused integration types (Slack, Teams, Datadog, New Relic, webhook) if they are not required.
Rotate any secrets referenced in Authorization, customHeaders, or DD-API-KEY fields if suspicious addon configurations are identified.

bash

# Example egress restriction using iptables on the Unleash host
iptables -A OUTPUT -d 127.0.0.0/8 -j REJECT
iptables -A OUTPUT -d 10.0.0.0/8 -j REJECT
iptables -A OUTPUT -d 172.16.0.0/12 -j REJECT
iptables -A OUTPUT -d 192.168.0.0/16 -j REJECT
iptables -A OUTPUT -d 169.254.0.0/16 -j REJECT