Description

Trigger.dev versions before 4.5.2 fail to validate environment membership during run replay operations, allowing authenticated attackers to inject task runs into arbitrary environments. Attackers can replay their own runs into other organizations' or projects' environments to consume victim resources and pollute run history.

Executive Summary

Technical Details

Mitigations

References

Additional Metadata

{
  "score": "8.5",
  "pubDate": "2026-09-04T15:17:43.347Z",
  "pubdate": "2026-09-04T15:17:43.347Z",
  "executiveSummary": "Trigger.dev versions prior to 4.5.2 are susceptible to an improper authorization vulnerability concerning run replay operations. The flaw stems from a failure to validate environment membership during the execution of a replay task. An authenticated attacker can exploit this logic error to inject task runs into arbitrary environments, including those belonging to separate organizations or projects. This unauthorized cross-environment execution poses significant security and operational risks, as attackers can force the consumption of victim-owned compute resources and manipulate run history to obfuscate malicious activities or disrupt legitimate workflows. Successful exploitation requires the attacker to be authenticated within the Trigger.dev ecosystem, but it does not require administrative privileges within the target environment. The vulnerability effectively allows for privilege escalation regarding environment interaction, undermining the logical isolation between isolated project workspaces.",
  "technicalDetails": "The vulnerability resides in the backend logic governing the 'replay' functionality for task runs within Trigger.dev. Specifically, the system fails to perform a rigorous authorization check to ensure the authenticated user possesses the appropriate permissions or membership for the target environment when initiating a replay request. In a secure implementation, the service should cross-reference the user's session identifier against the target environment's access control list (ACL) or organizational membership table before queuing a new task execution.\nThe exploitation flow begins with the attacker identifying a target environment identifier (environment_id) belonging to a different project or organization. Leveraging the existing 'replay' API endpoint, the attacker constructs a crafted request that references their own task run, but sets the destination environment to the victim's target identifier. Because the backend validates only the existence of the run, but fails to validate the ownership or cross-environment access rights during the transition to a new run execution, the server processes the request and injects the job into the victim's execution queue.\nAffected systems include Trigger.dev instances prior to version 4.5.2. This vulnerability is primarily classified as an Insecure Direct Object Reference (IDOR) variant or an authorization bypass that facilitates unauthorized resource interaction. The exploit can be initiated through standard authenticated network requests, meaning the attacker requires no special network position other than internet access to the application API.\nOnce the injected task runs in the victim's environment, the post-exploitation impact includes: 1) Resource Exhaustion: Triggering high-compute or long-running tasks within the victim's environment, leading to increased cloud costs or rate-limiting for the target organization. 2) Data Exfiltration/Manipulation: If the victim's environment is configured with sensitive credentials or API keys (e.g., environment variables) accessible to the triggered job, the injected task may leverage these to interact with third-party services. 3) Log Poisoning: The attacker can pollute the execution history of the victim's project, making forensic analysis, debugging, and auditing significantly more complex."
}