Skip to content
Threat Feed
high advisory

SQL Injection in Chatwoot FilterService

An authenticated SQL injection vulnerability (CVE-2026-44706) in Chatwoot versions 4.11.1 and earlier allows attackers to perform unauthorized database queries and exfiltrate sensitive data.

CVE search metadata

CVE search record: CVE-2026-44706. Severity: high. CVSS: 8.5. EPSS: 0.23%. KEV: no. Product: Chatwoot (<= 4.11.1). Brief: SQL Injection in Chatwoot FilterService. Brief link: https://feed.craftedsignal.io/briefs/2026-09-chatwoot-sqli/

Chatwoot versions 4.11.1 and earlier contain a critical SQL injection vulnerability in the FilterService, identified as CVE-2026-44706. The vulnerability resides in the FilterService#lt_gt_filter_values method, which fails to properly parameterize values within the is_greater_than operator used during conversation filtering. An authenticated attacker, such as a malicious or compromised agent, can send specially crafted requests to the /api/v1/accounts/{account_id}/conversations/filter endpoint to execute arbitrary SQL commands against the underlying PostgreSQL database. This allows for the exfiltration of sensitive information, including user credentials and API tokens. The availability of public proof-of-concept exploit code increases the risk of successful exploitation in enterprise environments.

Attack Chain

  1. Attacker obtains valid authentication credentials for a low-privileged agent account in the target Chatwoot instance.
  2. Attacker interacts with the /api/v1/accounts/{account_id}/conversations/filter API endpoint.
  3. Attacker injects malicious SQL payloads into the request body, specifically targeting the is_greater_than operator parameters.
  4. The vulnerable FilterService component processes the unparameterized input, executing the malicious SQL query on the PostgreSQL database.
  5. Attacker employs boolean-based or time-based blind SQL injection techniques to infer database structure and content.
  6. Attacker exfiltrates sensitive rows, specifically targeting user account tables, credentials, and API tokens.
  7. Final objective is achieved: unauthorized access to administrative credentials and internal platform data.

Impact

Successful exploitation of this vulnerability allows an authenticated agent to bypass standard authorization controls and dump the entire PostgreSQL database. This results in the exposure of highly sensitive data, including customer PII, internal communication logs, and administrative API tokens, which could be leveraged for lateral movement or full platform takeover.

Recommendation

  1. Upgrade all Chatwoot instances to a version later than 4.11.1 immediately.
  2. Implement strict monitoring on the /api/v1/accounts/{account_id}/conversations/filter endpoint for anomalous HTTP request bodies containing SQL syntax characters (e.g., ', --, UNION, SELECT).
  3. Review access logs for excessive or unusual activity originating from low-privileged agent accounts.
  4. Perform a security audit of current agent permissions to ensure the principle of least privilege is enforced within the Chatwoot dashboard.

Immediate actions

Upgrade Chatwoot to 4.11.2 or later

IT Operations 48h

Threat Hunt

Search logs for POST requests to conversation filter API containing SQL keywords

T1190 high medium confidence hunt now

Data: Web application request logs

Mitigations

Update Chatwoot to 4.11.2 or later

immediate IT Operations

CVE-2026-44706

Detection coverage 1

Detects CVE-2026-44706 Exploitation - SQL Injection via Chatwoot API

high

Detects exploitation of CVE-2026-44706 by monitoring for common SQL injection patterns in requests to the conversation filter endpoint

sigma tactics: initial_access techniques: T1190 sources: webserver

Detection queries are available on the platform. Get full rules →