Skyvern server-side request forgery in browser egress policy
Affects self-hosted Skyvern 1.0.56 using a remote browser (BROWSER_REMOTE_DEBUGGING_URL) on a different network from the worker. CVSS scores it Medium (5.1). The facts list no fixed version.
- Exploited
- No known exploitation
- Fix
- Fix or patch referenced
- Who is exposed
- Specific setups only
Summary
A server-side request forgery flaw in the classify_url function of the Skyvern-AI Skyvern browser egress policy lets an attacker make the server send requests to destinations of their choosing. It can be reached remotely by an attacker who needs low-privilege access and a user action to trigger it.
Written by a model
Original description
A security flaw has been discovered in Skyvern-AI Skyvern 1.0.56. The affected element is the function classify_url of the file skyvern/forge/sdk/browser_egress_policy.py. The manipulation results in server-side request forgery. The attack may be performed from remote. Upgrading to version 1.0.57 is sufficient to fix this issue. The patch is identified as 10b8d829c2750ebf62f2b70dec0e288297e752dd. It is recommended to upgrade the affected component.
Who should check
Teams running self-hosted Skyvern 1.0.56 or earlier with a remote browser (BROWSER_REMOTE_DEBUGGING_URL) on a network separate from the worker
“Affected: Skyvern-AI/skyvern at commit 623eb57 ( v1.0.56 )”
What to do
Check whether your setup matches what the bug needs.
Skyvern version 1.0.56 (the advisory's affected release) Self-hosted deployment using a remote browser (BROWSER_REMOTE_DEBUGGING_URL) Browser runs on a different network from the worker, so the browser can resolve a hostname the worker cannotApply the patch listed under Evidence.
Check your systems with an agent
Paste this into a coding or ops agent. It checks your systems read-only and reports a table per asset.
Check whether my systems are affected by CVE-2026-134534 (Skyvern-AI Skyvern browser_egress_policy.py classify_url server-side request forgery), using the cve.now data below.
Rules: work read-only. Do not run exploit code, proof-of-concept scripts or scanners that send attack payloads, and do not change any configuration, package, image or deployment. Report findings only.
What cve.now has for CVE-2026-134534 (entry last updated 2026-11-21):
Affected:
- Skyvern-AI Skyvern: versions 1.0.56
Fixed versions: none listed yet.
Conditions (all must be true for the bug to be hit):
- Skyvern version 1.0.56 (the advisory's affected release) (source: https://github.com/Skyvern-AI/skyvern/issues/9010)
- Self-hosted deployment using a remote browser (BROWSER_REMOTE_DEBUGGING_URL) (source: https://github.com/Skyvern-AI/skyvern/issues/9010)
- Browser runs on a different network from the worker, so the browser can resolve a hostname the worker cannot (source: https://github.com/Skyvern-AI/skyvern/issues/9010)
Setup: the bug needs a feature or setting that is off by default or uncommon. Check whether each asset has it on.
Attacker needs: network access, low-privilege login, a user action.
CISA KEV: CVE-2026-134534 is not listed as of cve.now's last read.
This KEV status is for CVE-2026-134534 only. If you also check another CVE, look up its own KEV status; never assume it from this one.
How to check:
- Installed software: check what is actually running, not what a repo declares. Read the version from each host, container image and Kubernetes node (the binary's version output, the OS package database, image layers), and check whether the conditions above hold there. Distribution packages may carry a backported fix under an older version number; confirm against the distribution's own advisory.
Verify against the linked advisories before concluding, and cite the date you read each one:
- https://github.com/Skyvern-AI/skyvern/pull/9013
- https://github.com/Skyvern-AI/skyvern/commit/10b8d829c2750ebf62f2b70dec0e288297e752dd
- https://github.com/Skyvern-AI/skyvern/releases/tag/v1.0.57
- https://github.com/Skyvern-AI/skyvern/issues/9010
- https://vuldb.com/vuln/441679
- https://vuldb.com/vuln/441679/cti
- https://vuldb.com/cve/CVE-2026-134534
- https://vuldb.com/submit/1052272
Do not conclude an asset is unaffected unless you checked it and the evidence covers both its version and the conditions above. If you could not check something, mark it unknown and say what is missing.
Report a table with one row per asset and these columns: Asset | Version or build | Status (affected, fixed, not applicable, unknown) | Compensating controls | Exposure (internet-facing, internal, none) | Evidence | Next action.
Entry: https://cve.now/CVE-2026-134534
For follow-up questions, use the cve.now MCP server at https://mcp.cve.now/mcp (no authentication), for example execute with name cve.get and arguments {"id": "CVE-2026-134534"}.Affected versions
- Skyvern-AI Skyvern1.0.56
cve.now Index
- Exploitation?/40Unknown. No CISA SSVC exploitation rating yet
- Exposure8/25Specific setups only, reachable over the network
- Reach?/20Unknown. No tracked technology or package matched
- Severity5.1/10CVSS 5.1
- No patch0/5No signal
What would change the score
- Would score about 63 if added to CISA KEV
- Would score about 58 if CISA's SSVC rates exploitation active
- Would score about 20 if CISA's SSVC rates it automatable
Timeline
- CVE published
- First seen by cve.now
Evidence 11
| Kind | Source |
|---|---|
| CVE record 2Establishes: The identifier and description. | |
| Record | CVE List · CVE-2026-134534 |
| Record | NVD · CVE-2026-134534 |
| Other links 9 | |
| Advisory | VulDB · Advisory #441679 |
| Advisory | VulDB · Threat intelligence #441679 |
| Advisory | VulDB · CVE-2026-134534 record |
| Advisory | VulDB · Submission #1052272 |
| Patch | GitHub · Skyvern-AI/skyvern · Pull request #9013 |
| Patch | GitHub · Skyvern-AI/skyvern · Commit 10b8d829 |
| Patch | GitHub · Skyvern-AI/skyvern · Release v1.0.57 |
| Report | GitHub · Skyvern-AI/skyvern · Issue #9010 |
| Repository | GitHub · Skyvern-AI/skyvern · Repository |
Questions about CVE-2026-134534
Is CVE-2026-134534 being exploited?
It is not in CISA's Known Exploited Vulnerabilities catalog, and CISA's SSVC assessment does not mark exploitation as active. That does not mean it has not been attacked: vendors and researchers can report attacks before CISA lists them. No public exploit code is tracked.
Source: cve.now feed data
What does this mean?
The URL classification in browser_egress_policy.py can be bypassed, so requests reach internal HTTP(S) services that the browser's network can reach. The browser can resolve a hostname the worker cannot, and the agent renders and extracts the response, so internal content can be read and leave through the run's outbound channels.
Source: github.com
Which versions are affected by CVE-2026-134534?
Skyvern: version 1.0.56.
Source: cve.now feed data
Am I affected, and how do I check?
The advisory names Skyvern 1.0.56 (commit 623eb57, tag v1.0.56). The exposed setup is a self-hosted deployment that uses a remote browser through BROWSER_REMOTE_DEBUGGING_URL, with the browser on a different network from the worker. To check a setup, look at the installed Skyvern version and whether BROWSER_REMOTE_DEBUGGING_URL is set.
Source: github.com
How do I fix CVE-2026-134534?
The facts do not list a fixed version or vendor patch ids for this vulnerability, so no upgrade target is named. Check the vendor's advisory for an update before relying on any release.
Source: github.com
Related
- CVE-2026-123783Skyvern through 1.0.56 Script Sandbox Bypass via types.CodeType Bytecode
- CVE-2026-82447Skyvern before 1.0.45 Sandbox Escape via TextPromptBlock
- CVE-2026-134533Skyvern-AI Skyvern browser_egress_policy.py classify_url server-side request forgery
- CVE-2026-130235Improper Request URL Validation in the Okta .NET SDK Abstractions Library
- CVE-2026-77315Flyto2 Core: http.paginate follows server-supplied next-page URLs (Link header / cursor) without SSRF revalidation
- CVE-2026-129029WordPress OttoKit plugin <= 1.1.42 - Server Side Request Forgery (SSRF) vulnerability
No corrections so far. Something wrong on this page? Report an error.