GHSA-xg5g-26x8-cvf4 is a high-severity (CVSS 8.5) Server-Side Request Forgery (SSRF) vulnerability in @budibase/server. O3 Security confirms whether GHSA-xg5g-26x8-cvf4 is actually reachable in your code before you act, and blocks exploitation at runtime until you patch.
Budibase: DNS rebinding SSRF bypasses remain in OpenAPI import and REST query execution
Exploitation Status
Proof-of-concept exploit code exists
- CISA’s SSVC triage found public proof-of-concept exploit code for this CVE, though no confirmed active exploitation.
Exploitation and automatability from CISA’s SSVC triage for GHSA-xg5g-26x8-cvf4.
EPSS Exploitation Probability
EPSS (Exploit Prediction Scoring System) is a daily probability model maintained by FIRST.org. It estimates the likelihood a CVE will be exploited in production environments within the next 30 days, derived from real-world threat intelligence signals.
How urgent is this, really
GHSA-xg5g-26x8-cvf4 plotted by exploitation likelihood (EPSS) against impact (CVSS). The shaded corner — EPSS 50%+ and CVSS 7.0+ — is where this CVE doesn't sit, though severity or exploitability alone can still warrant action.
Where this sits among everything scored
Of 369,023 CVEs with a current EPSS score, this one falls in the < 10% band (highlighted). Real counts from FIRST.org, not a sample — log-scaled since the landscape is heavily right-skewed.
Real-World Exposure
How broadly this vulnerability is actually deployed: weekly install volume shows current usage, and reverse-dependency count shows how many other packages break if it stays unpatched.
@budibase/servernpmDescription
Impact
A builder-level user can make Budibase issue server-side HTTP requests to loopback or private-network targets by using DNS rebinding against two outbound fetch paths that are still not pinned to the validated DNS answer.
The first path is OpenAPI query import. It validates the supplied hostname with the blacklist and then performs a separate raw fetch. A hostname that resolves to a public address during validation and to 127.0.0.1 during the request is accepted.
The second path is REST datasource/query execution. It calls the fixed fetchWithBlacklist() helper, but provides a custom undici fetchFn that installs a dispatcher. The helper passes a pinned Node agent, but undici uses the dispatcher and resolves the original hostname again at connection time. This defeats the DNS pinning added for the previous outbound fetch advisory.
Impact is authenticated SSRF from the Budibase server process. An attacker with the relevant builder/query privileges can reach loopback or private HTTP services that should be blocked by Budibase's outbound fetch protections, subject to the normal response handling of each feature.
Reproduction
- Configure an attacker-controlled hostname so the validation lookup returns a public IP address and the later request lookup returns an internal address such as
127.0.0.1. - Start a loopback canary HTTP service on the Budibase host.
- For OpenAPI query import, submit an import URL such as
http://rebind.test:<port>/openapi.jsonthrough the builder query import endpoints. In a harness matching the current import code, validation resolvedrebind.testto8.8.8.8, the raw fetch then resolvedrebind.testto127.0.0.1, and the loopback canary returnedloopback-canary-hit:/openapi.json. - For REST datasource/query execution, configure a REST datasource/query URL using the same rebinding hostname. In a harness matching the current helper plus REST undici custom fetch path, a direct loopback URL was blocked and a pinned Node-agent control did not hit loopback, but the REST undici path re-resolved the hostname to
127.0.0.1and returnedREST-UNDICI-REBIND-CANARYfrom the loopback service.
Observed OpenAPI import proof:
{
"outcome": "allowed",
"body": "loopback-canary-hit:/openapi.json",
"lookups": [
{"phase": "guard", "hostname": "rebind.test", "address": "8.8.8.8"},
{"phase": "fetch", "hostname": "rebind.test", "address": "127.0.0.1"}
]
}
Observed REST undici proof:
{
"directControl": {"outcome": "blocked", "hitsAfter": 0},
"pinnedNodeFetchControl": {"outcome": "no_loopback_hit", "hitsAfter": 0},
"undiciBypass": {
"outcome": "allowed",
"status": 200,
"body": "REST-UNDICI-REBIND-CANARY",
"hitsAfter": 1
},
"lookupTrace": [
{"phase": "guard", "hostname": "rebind.test", "address": "8.8.8.8"},
{"phase": "fetch", "hostname": "rebind.test", "address": "127.0.0.1"}
]
}
Root cause and technical details
The central helper now resolves and validates the target hostname, then passes a pinned Node http.Agent or https.Agent to the request. That protects callers whose HTTP client honors the Node agent option.
OpenAPI query import does not use that helper. In packages/server/src/api/controllers/query/import/index.ts, assertUrlIsSafe() calls blacklist.isBlacklisted(parsed.hostname), but fetchFromUrl() then calls fetch(currentUrl, { redirect: "manual" }). That creates a validate-then-fetch gap where DNS can change between the blacklist lookup and the actual connection.
REST datasource/query execution uses the helper but changes the transport semantics. In packages/server/src/integrations/rest.ts, setDispatcher() creates an undici dispatcher with getDispatcher({ rejectUnauthorized, url: requestUrl }), then the custom fetch function calls fetch(requestUrl, setDispatcher(requestInput, requestUrl)). In packages/backend-core/src/utils/outboundFetch.ts, fetchWithBlacklist() passes the pinned Node agent to fetchFn, but undici fetch uses the supplied dispatcher instead. The actual connection performs a second hostname resolution and can be rebound to a blocked address.
Current affected code is present in latest master and in the latest release tag 3.39.14.
Remediation
Route OpenAPI query import through the pinned outbound fetch helper rather than validating and raw-fetching separately.
Make fetchWithBlacklist() pin the actual connection for every supported transport. For undici callers, provide a dispatcher whose lookup or connect behavior is pinned to the validated IP, or reject custom fetch functions that do not explicitly enforce the pinned address. Add regression coverage for direct loopback blocking, DNS rebinding through OpenAPI import, DNS rebinding through REST undici dispatcher, redirect-to-loopback, IPv4-mapped IPv6, and mixed public/private DNS answers.
Affected Packages
| Ecosystem | Package | Vulnerable range | Fix |
|---|---|---|---|
| 📦npm | @budibase/server | all versions | No fix |
Detection & mitigation playbook
Open-source dependencyDetect
Scan your dependency tree (package-lock.json, pnpm-lock.yaml, requirements.txt, go.sum, etc.) for @budibase/server. O3's reachability analysis confirms whether the vulnerable code path is actually invoked in your application, so you act on real exposure instead of every transitive match.
Remediation status
No patched version of @budibase/server has shipped for GHSA-xg5g-26x8-cvf4 yet. Where your build allows, override or pin the dependency away from the vulnerable range, and apply any maintainer-recommended mitigation.
Mitigate without a patch
If you can't upgrade right away: gate or disable the affected feature, validate untrusted input at the boundary, and avoid passing attacker-controlled data into the vulnerable path. O3's runtime protection blocks exploitation in production as an interim safeguard until the upgrade lands.
How O3 protects you
O3 pinpoints whether GHSA-xg5g-26x8-cvf4 is reachable in your code and exactly where to fix it, then blocks exploitation in production at runtime until the patched version is deployed.
Tailored to GHSA-xg5g-26x8-cvf4. Runtime protection reduces exposure until a permanent patch is applied and verified — it complements patching, it doesn't replace it.
Frequently Asked Questions
Is GHSA-xg5g-26x8-cvf4 in your dependencies?
O3 detects GHSA-xg5g-26x8-cvf4 across npm dependencies and uses function-level reachability to confirm whether the vulnerable code path is actually reachable — not just present. No false positives.