Description

pgweb through 0.17.0 leaves the POST /api/connect endpoint unguarded when connect-backend authorization is configured, allowing attackers to supply arbitrary database connection strings. Attackers can bypass the resource-to-database mapping by providing a custom session identifier and connection URL to access unauthorized databases and internal services.

Executive Summary

Technical Details

Mitigations

References

Additional Metadata

{
  "score": "8.5",
  "pubDate": "2026-09-15T11:17:13.053Z",
  "pubdate": "2026-09-15T11:17:13.053Z",
  "executiveSummary": "The pgweb application, through version 0.17.0, contains a critical authentication bypass vulnerability within its connect-backend implementation. Specifically, the POST /api/connect endpoint fails to properly validate requests when authorization is configured, enabling remote attackers to perform arbitrary database connection string injection. This flaw allows unauthorized actors to bypass established resource-to-database mapping constraints and interact with internal database services or unauthorized external targets. The vulnerability poses significant risk, as it effectively nullifies backend access controls, granting attackers the ability to manipulate connection parameters for unauthorized data access or service probing. Successful exploitation requires no prior authentication and can be performed via standard HTTP POST requests against the vulnerable API endpoint. The impact encompasses potential unauthorized data exfiltration, service discovery, and potential lateral movement within the infrastructure hosting the pgweb service.",
  "technicalDetails": "The root cause of this vulnerability lies in an improper authorization check within the pgweb backend logic associated with the /api/connect endpoint. While pgweb supports a connect-backend authorization configuration intended to restrict user access to pre-defined database resources, the implementation fails to enforce these restrictions when the application processes connection requests. Specifically, the backend architecture incorrectly assumes that incoming POST requests to /api/connect are implicitly authorized or that the mapping validation logic is consistently applied across all request flows.\nThe exploitation flow begins with the attacker targeting the /api/connect endpoint. By constructing a malicious POST request, the attacker can override the intended session-based resource mapping by supplying a custom session identifier and a crafted connection URL. Because the application logic does not validate the provided connection string against the authenticated session’s permitted scope, the backend processes the attacker-supplied URI as a legitimate connection request. This allows the attacker to connect to any reachable database instance—internal or external—that the server process has network visibility toward.\nTechnically, the vulnerability exists because the server-side code fails to verify the integrity of the connection parameters provided by the client during the session lifecycle. The application treats the user-provided connection string as trusted input rather than validating it against a server-side whitelist or a pre-configured resource map. Consequently, an attacker can bypass the intended multi-tenancy or access control boundaries defined in the pgweb configuration. Once the connection is established, the attacker gains the ability to execute arbitrary SQL commands on the target database, effectively assuming the privileges of the database user account defined in their injected connection string. Post-exploitation activities are limited only by the permissions of the database user account used during the connection and the network reachability of the pgweb host. This represents a severe failure in security-critical code paths responsible for backend access enforcement in all versions up to and including 0.17.0."
}