GHSA-VG6P-V9VM-6FGJ
Vulnerability from github – Published: 2026-08-25 14:43 – Updated: 2026-08-25 14:43The web_crawl tool performs its SSRF check only on the initial URL: it resolves the hostname once with socket.gethostbyname and rejects private/loopback/link-local results. It then passes the URL to a fetcher that uses httpx.Client(follow_redirects=True) - or urllib.request.urlopen when httpx is absent, which also follows redirects - and re-resolves the hostname at connect time, with no further validation. This validate-here/fetch-there gap is bypassable two independent ways: HTTP redirects and DNS rebinding.
Affected code: src/praisonai-agents/praisonaiagents/tools/web_crawl_tools.py - Single-shot validation (lines 229-238): if os.environ.get("ALLOW_LOCAL_CRAWL") != "true": ip_str = socket.gethostbyname(hostname) # resolved ONCE, at validation time ip = ipaddress.ip_address(ip_str) if ip.is_loopback or ip.is_private or ip.is_link_local or ip.is_multicast or ip.is_unspecified: continue # rejected url_list.append(u) - Vulnerable fetch (_crawl_with_httpx, lines 142 / 149): follows redirects, re-resolves DNS, no re-check: with httpx.Client(follow_redirects=True, timeout=30.0) as client: response = client.get(url) # fallback: urllib.request.urlopen(url, timeout=30) (also follows redirects by default) - web_crawl / crawl_web are registered tools (tools/init.py:156-157); httpx is the default fallback provider (dispatch at web_crawl_tools.py:269).
The two bypasses: 1) Redirect: validation approves an attacker domain resolving to a public IP; attacker server replies 302 Location: http://169.254.169.254/... (or any internal host); the fetcher follows it unchecked. 2) DNS rebinding (TOCTOU): validator's gethostbyname and fetcher's connect-time resolution are independent; a low-TTL attacker domain answers public to the validator and private/loopback to fetch.
Impact: An agent with web_crawl - driven by direct input or indirect prompt injection - can be made to read internal-only HTTP services and cloud instance-metadata endpoints (e.g. IAM credentials), with the response body returned in the tool output. Scope is Changed because the request pivots into the internal network.
Proof of concept: A PoC drives the real web_crawl() (httpx absent -> genuine urllib fallback). It runs a loopback "internal metadata" service and a loopback attacker redirector, substituting DNS only to stand in for "attacker owns a public domain" / offline routing - the redirect-following and connect-time re-resolution are the repo's own behavior. Observed: CONTROL: web_crawl("http://127.0.0.1:.../meta-data/") -> blocked (validator works) PoC 1A (redirect): attacker.example approved (public); 302 -> loopback metadata -> result.content leaks {"AccessKeyId":"ASIA_FAKE_STOLEN_CREDENTIAL_..."} PoC 1B (rebinding): gethostbyname(rebind.example)->public (allowed); connect->127.0.0.1 -> same secret leaked The control proves the validator blocks a direct loopback request, so the bypasses are genuine.
Remediation: Resolve the hostname once, validate that IP, and connect to that exact validated IP (pin it) rather than re-resolving. Disable redirect following (follow_redirects=False; for urllib use a redirect handler that re-validates), or re-validate every redirect hop's resolved IP. Apply the deny check to both the validator and the actual socket target. file_tools.py:364 already uses follow_redirects=False and is the correct pattern to propagate.
Distinct from prior advisories: The accepted SSRF advisories concern host-string parsing in different code — alternate loopback encodings in spider_tools (GHSA-5c6w-wwfq-7qqm) and the CLI @url feature (GHSA-5cxw-77wg-jrf3). This is in the web_crawl tool, which neither advisory names, and the mechanisms (redirect-following and DNS rebinding) differ categorically from host-string encoding; the spider_tools _host_is_blocked hardening does not apply to this tool.
{
"affected": [
{
"package": {
"ecosystem": "PyPI",
"name": "praisonaiagents"
},
"ranges": [
{
"events": [
{
"introduced": "0"
},
{
"fixed": "1.6.58"
}
],
"type": "ECOSYSTEM"
}
]
}
],
"aliases": [
"CVE-2026-55524"
],
"database_specific": {
"cwe_ids": [
"CWE-367",
"CWE-918"
],
"github_reviewed": true,
"github_reviewed_at": "2026-08-25T14:43:19Z",
"nvd_published_at": "2026-08-05T20:17:10Z",
"severity": "HIGH"
},
"details": "The web_crawl tool performs its SSRF check only on the initial URL: it resolves the hostname once\nwith socket.gethostbyname and rejects private/loopback/link-local results. It then passes the URL to\na fetcher that uses httpx.Client(follow_redirects=True) - or urllib.request.urlopen when httpx is\nabsent, which also follows redirects - and re-resolves the hostname at connect time, with no further\nvalidation. This validate-here/fetch-there gap is bypassable two independent ways: HTTP redirects and\nDNS rebinding.\n\nAffected code: src/praisonai-agents/praisonaiagents/tools/web_crawl_tools.py\n- Single-shot validation (lines 229-238):\n if os.environ.get(\"ALLOW_LOCAL_CRAWL\") != \"true\":\n ip_str = socket.gethostbyname(hostname) # resolved ONCE, at validation time\n ip = ipaddress.ip_address(ip_str)\n if ip.is_loopback or ip.is_private or ip.is_link_local or ip.is_multicast or ip.is_unspecified:\n continue # rejected\n url_list.append(u)\n- Vulnerable fetch (_crawl_with_httpx, lines 142 / 149): follows redirects, re-resolves DNS, no re-check:\n with httpx.Client(follow_redirects=True, timeout=30.0) as client: response = client.get(url)\n # fallback: urllib.request.urlopen(url, timeout=30) (also follows redirects by default)\n- web_crawl / crawl_web are registered tools (tools/__init__.py:156-157); httpx is the default fallback\n provider (dispatch at web_crawl_tools.py:269).\n\nThe two bypasses:\n1) Redirect: validation approves an attacker domain resolving to a public IP; attacker server replies\n 302 Location: http://169.254.169.254/... (or any internal host); the fetcher follows it unchecked.\n2) DNS rebinding (TOCTOU): validator\u0027s gethostbyname and fetcher\u0027s connect-time resolution are\n independent; a low-TTL attacker domain answers public to the validator and private/loopback to fetch.\n\nImpact:\nAn agent with web_crawl - driven by direct input or indirect prompt injection - can be made to read\ninternal-only HTTP services and cloud instance-metadata endpoints (e.g. IAM credentials), with the\nresponse body returned in the tool output. Scope is Changed because the request pivots into the\ninternal network.\n\nProof of concept:\nA PoC drives the real web_crawl() (httpx absent -\u003e genuine urllib fallback). It runs a loopback\n\"internal metadata\" service and a loopback attacker redirector, substituting DNS only to stand in\nfor \"attacker owns a public domain\" / offline routing - the redirect-following and connect-time\nre-resolution are the repo\u0027s own behavior. Observed:\n CONTROL: web_crawl(\"http://127.0.0.1:.../meta-data/\") -\u003e blocked (validator works)\n PoC 1A (redirect): attacker.example approved (public); 302 -\u003e loopback metadata\n -\u003e result.content leaks {\"AccessKeyId\":\"ASIA_FAKE_STOLEN_CREDENTIAL_...\"}\n PoC 1B (rebinding): gethostbyname(rebind.example)-\u003epublic (allowed); connect-\u003e127.0.0.1\n -\u003e same secret leaked\nThe control proves the validator blocks a direct loopback request, so the bypasses are genuine.\n\nRemediation:\nResolve the hostname once, validate that IP, and connect to that exact validated IP (pin it) rather\nthan re-resolving. Disable redirect following (follow_redirects=False; for urllib use a redirect\nhandler that re-validates), or re-validate every redirect hop\u0027s resolved IP. Apply the deny check to\nboth the validator and the actual socket target. file_tools.py:364 already uses follow_redirects=False\nand is the correct pattern to propagate.\n\nDistinct from prior advisories:\nThe accepted SSRF advisories concern host-string parsing in different code \u2014 alternate loopback\nencodings in spider_tools (GHSA-5c6w-wwfq-7qqm) and the CLI @url feature (GHSA-5cxw-77wg-jrf3). This\nis in the web_crawl tool, which neither advisory names, and the mechanisms (redirect-following and DNS\nrebinding) differ categorically from host-string encoding; the spider_tools _host_is_blocked hardening\ndoes not apply to this tool.",
"id": "GHSA-vg6p-v9vm-6fgj",
"modified": "2026-08-25T14:43:19Z",
"published": "2026-08-25T14:43:19Z",
"references": [
{
"type": "WEB",
"url": "https://github.com/MervinPraison/PraisonAI/security/advisories/GHSA-vg6p-v9vm-6fgj"
},
{
"type": "ADVISORY",
"url": "https://nvd.nist.gov/vuln/detail/CVE-2026-55524"
},
{
"type": "WEB",
"url": "https://github.com/MervinPraison/PraisonAI/commit/2f9677abb2ea68eab864ee8b6a828fd0141612e1"
},
{
"type": "PACKAGE",
"url": "https://github.com/MervinPraison/PraisonAI"
},
{
"type": "WEB",
"url": "https://github.com/MervinPraison/PraisonAI/releases/tag/v4.6.58"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:3.1/AV:N/AC:H/PR:N/UI:N/S:C/C:H/I:L/A:N",
"type": "CVSS_V3"
}
],
"summary": "praisonaiagents vulnerable to SSRF in web_crawl tool via redirect-following and DNS rebinding (validate-then-fetch gap)"
}
Sightings
| Author | Source | Type | Date | Other |
|---|
Nomenclature
- Seen: The vulnerability was mentioned, discussed, or observed by the user.
- Confirmed: The vulnerability has been validated from an analyst's perspective.
- Published Proof of Concept: A public proof of concept is available for this vulnerability.
- Exploited: The vulnerability was observed as exploited by the user who reported the sighting.
- Patched: The vulnerability was observed as successfully patched by the user who reported the sighting.
- Not exploited: The vulnerability was not observed as exploited by the user who reported the sighting.
- Not confirmed: The user expressed doubt about the validity of the vulnerability.
- Not patched: The vulnerability was not observed as successfully patched by the user who reported the sighting.
The approach is described in our paper Mapping CVEs to MITRE ATT&CK Techniques: A Curated Gold-Set Classifier and the Limits of LLM-Assisted Label Expansion.
Browse all ATT&CK techniques and the vulnerabilities related to each.
Related by attack behaviour
Vulnerabilities whose description is nearest to this one in the vector space of the CIRCL/vulnerability-attack-technique-biencoder model. This is a similarity search over the bi-encoder space (plain cosine), not a classification, and it has no measured accuracy.