CVE Intelligence
Analysis AI-assisted

CVE-2025-62615: AutoGPT Platform RSSFeedBlock SSRF

CVE-2025-62615 is a server-side request forgery (SSRF) in the RSSFeedBlock of AutoGPT Platform, where the URL passed to urllib.request.urlopen is not filtered. NVD rates it 9.8 (critical) under CVSS 3.1, and the project states it was patched in autogpt-platform-beta-v0.6.34. The record shows no KEV listing, and no confirmed exploitation appears in the sources consulted.

Published Oct 13, 2026 · Updated Oct 13, 2026 · Ler em português

Overview

CVE-2025-62615 affects the RSSFeedBlock component of AutoGPT, a platform for building and running continuous AI agents that automate workflows. According to the GitHub advisory, the block calls urllib.request.urlopen directly on a URL supplied as input without filtering it. This is classified as CWE-918, Server-Side Request Forgery. The issue was patched in autogpt-platform-beta-v0.6.34 (GitHub advisory).

Impact

SSRF allows a request-making component to be induced to send requests to destinations chosen by the input. Depending on the deployment, this could expose internal services or cloud metadata endpoints that are reachable from the platform host. NVD assigns a CVSS 3.1 base score of 9.8 with vector CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H, and the CNA GitHub reports a CVSS 4.0 score of 9.3. The vector indicates network reachability, low complexity, no privileges and no user interaction, with high impact to confidentiality, integrity and availability.

Affected products and versions

  • Product listed as Agpt Autogpt Platform, versions < 0.6.34.
  • The description refers to autogpt-platform-beta-v0.6.34 as the first patched release.
  • The record lists no separate fixed-version field, so the fix version is taken from the description and the advisory.

Exploitation status

The CISA KEV catalog does not list this CVE in the record. No confirmed active exploitation appears in the sources consulted. One reference is tagged as Exploit in the record, but that tag alone does not confirm exploitation in the wild, and this analysis does not describe any exploit material.

Remediation and mitigation

  • Upgrade AutoGPT Platform to autogpt-platform-beta-v0.6.34 or a later release, as stated by the project.
  • Until patched, review which workflows use the RSSFeedBlock and limit who can create or edit them.
  • Restrict outbound traffic from the platform host with egress allowlists, so it cannot reach internal networks, management interfaces or cloud metadata endpoints unnecessarily.
  • Apply network segmentation so that the AutoGPT host has minimal access to sensitive internal systems.

Detection guidance

  • Review platform logs for workflows that fetch feeds from unusual or internal addresses.
  • Monitor egress network logs from the AutoGPT host for connections to private address ranges or metadata services that are not expected.
  • Track changes to RSSFeedBlock workflow definitions and alert on newly created feed inputs.

Limitations of the available information

The record does not include a full list of affected builds beyond the version range, EPSS data, SSVC data or independent proof-of-concept analysis. The CVSS 4.0 score comes from the CNA record, while the CVSS 3.1 score is from NVD. Whether a given deployment is exposed depends on its configuration, which the sources do not describe.

Frequently asked questions

What is CVE-2025-62615?

It is a server-side request forgery vulnerability in the RSSFeedBlock of AutoGPT Platform, where the input URL is not filtered before being opened with urllib.request.urlopen.

Which version fixes it?

The project states the issue is patched in autogpt-platform-beta-v0.6.34, and the affected range in the record is versions below 0.6.34.

This analysis was drafted with AI assistance from verified source data and automatically checked against it.