PYSEC-2026-3916
Vulnerability from pysec - Published: 2026-09-10 09:44 - Updated: 2026-09-10 11:02Summary
The Reachy Mini daemon exposes the “/api/media/sounds/upload” endpoint without authentication and file validation mechanisms.
An attacker can use this endpoint to upload malicious files into the file system that will propagate in future attacks.
Compromise Chain: Unauthenticated to Full Root Access
This issue is part of a full compromise chain allowing an unauthenticated user to gain root access on the Reachy’s operating system:
- Unrestricted File Upload in Media Sounds Upload API \<= current finding
- Bluetooth Authentication Bypass
- Bluetooth Directory Traversal
Description
The root cause of the issue is at the handler located in “src/daemon/app/routers/media.py” file at the “upload_sound” method:
@router.post("/sounds/upload")
async def upload_sound(
file: UploadFile = File(...),
) -> dict[str, str]:
"""Upload a sound file to the daemon's temporary sound directory.
The file is saved to ``/tmp/reachy_mini_sounds/<original_filename>``.
If a file with the same name already exists it is overwritten.
Returns:
JSON with the absolute *path* of the saved file on the daemon.
"""
if not file.filename:
raise HTTPException(status_code=400, detail="Filename is required")
# Reject path traversal
filename = Path(file.filename).name
if not filename or filename in (".", ".."):
raise HTTPException(status_code=400, detail="Invalid filename")
os.makedirs(SOUNDS_TMP_DIR, exist_ok=True)
dest = os.path.join(SOUNDS_TMP_DIR, filename)
content = await file.read()
with open(dest, "wb") as f:
f.write(content)
return {"status": "ok", "path": dest}
This endpoint lacks multiple defence mechanisms:
- No authentication mechanism.
- No file extension validation.
- No file content/size validation.
Additionally, the daemon is bound to the 0.0.0.0 network interfaces (a.k.a. all network interfaces) by default along with permissive CORS ( allow_origins=[“*”] ) meaning the following API endpoint is exposed to every network interface the daemon is connected to.
PoC
- Start the daemon in simulation mode (command depends on the installed environment):
.venv/bin/mjpython -m reachy_mini.daemon.app.main --sim --no-media
- After that check that the media upload API endpoint is activated and you can upload a wav file:
curl -X POST http://<daemon_domain>:<daemon_port>/api/media/sounds/upload \
-F "file=@/path/to/your/file.wav"
- Now attempt to create a “.sh” file containing a script and upload it:
curl -X POST http://<daemon_domain>:<daemon_port>/api/media/sounds/upload \
-F "file=@/path/to/your/script.sh"
- Now error message will be received and you will see that the script file was successfully uploaded to disk.
Impact
Due to this issue, an attacker can upload malicious files instead of the intended sounds files, harming the integrity of the stored data and allowing an attacker to propagate a foothold in cases another vulnerabilities would arise.
Fix suggestion
Perform the following check on the API endpoint:
- Validate that the file extension contains only desired extensions (allow-list approach).
- Validate that the uploaded file’s content matches the desired extension (Magic numbers, and known file structure per file type).
- Enforce authentication on the file upload endpoint.
Credit
The vulnerability was discovered by Natan Nehorai of the JFrog Vulnerability Research team.
| Name | purl | reachy-mini | pkg:pypi/reachy-mini |
|---|
{
"affected": [
{
"package": {
"ecosystem": "PyPI",
"name": "reachy-mini",
"purl": "pkg:pypi/reachy-mini"
},
"ranges": [
{
"events": [
{
"introduced": "0"
},
{
"fixed": "1.8.2"
}
],
"type": "ECOSYSTEM"
}
],
"versions": [
"1.0.0",
"1.0.0rc1",
"1.0.0rc4",
"1.0.0rc5",
"1.1.0",
"1.1.0rc3",
"1.1.0rc4",
"1.1.1",
"1.1.2",
"1.1.3",
"1.1.4",
"1.2.0",
"1.2.0rc1",
"1.2.1",
"1.2.10",
"1.2.11",
"1.2.13",
"1.2.2",
"1.2.3",
"1.2.3rc1",
"1.2.4",
"1.2.4rc1",
"1.2.4rc2",
"1.2.4rc3",
"1.2.4rc4",
"1.2.5",
"1.2.5rc1",
"1.2.5rc2",
"1.2.5rc3",
"1.2.5rc4",
"1.2.6",
"1.2.6rc1",
"1.2.6rc2",
"1.2.6rc3",
"1.2.7",
"1.2.7rc1",
"1.2.7rc2",
"1.2.8",
"1.2.9",
"1.3.0",
"1.3.1",
"1.4.0",
"1.4.0.dev0",
"1.4.1",
"1.5.0",
"1.5.1",
"1.6.0",
"1.6.1",
"1.6.2",
"1.6.3",
"1.6.4",
"1.7.0",
"1.7.1",
"1.7.2",
"1.7.3",
"1.8.0",
"1.8.0rc1",
"1.8.1"
]
}
],
"aliases": [
"CVE-2026-55419",
"GHSA-m2pc-3q4q-w6jr"
],
"details": "## Summary\n\nThe Reachy Mini daemon exposes the \u201c/api/media/sounds/upload\u201d endpoint without authentication and file validation mechanisms. \nAn attacker can use this endpoint to upload malicious files into the file system that will propagate in future attacks.\n\n## Compromise Chain: Unauthenticated to Full Root Access\n\nThis issue is part of a full compromise chain allowing an unauthenticated user to gain root access on the Reachy\u2019s operating system:\n\n1. Unrestricted File Upload in Media Sounds Upload API \\\u003c= current finding \n2. Bluetooth Authentication Bypass \n3. Bluetooth Directory Traversal\n\n\n## Description\n\nThe root cause of the issue is at the handler located in \u201c***src/daemon/app/routers/media.py***\u201d file at the \u201cupload\\_sound\u201d method:\n\n```py\n@router.post(\"/sounds/upload\")\nasync def upload_sound(\n file: UploadFile = File(...),\n) -\u003e dict[str, str]:\n \"\"\"Upload a sound file to the daemon\u0027s temporary sound directory.\n The file is saved to ``/tmp/reachy_mini_sounds/\u003coriginal_filename\u003e``.\n If a file with the same name already exists it is overwritten.\n Returns:\n JSON with the absolute *path* of the saved file on the daemon.\n \"\"\"\n if not file.filename:\n raise HTTPException(status_code=400, detail=\"Filename is required\")\n # Reject path traversal\n filename = Path(file.filename).name\n if not filename or filename in (\".\", \"..\"):\n raise HTTPException(status_code=400, detail=\"Invalid filename\")\n os.makedirs(SOUNDS_TMP_DIR, exist_ok=True)\n dest = os.path.join(SOUNDS_TMP_DIR, filename)\n content = await file.read()\n with open(dest, \"wb\") as f:\n f.write(content)\n return {\"status\": \"ok\", \"path\": dest}\n```\n\nThis endpoint lacks multiple defence mechanisms:\n\n1. No authentication mechanism. \n2. No file extension validation. \n3. No file content/size validation.\n\nAdditionally, the daemon is bound to the 0.0.0.0 network interfaces (a.k.a. all network interfaces) by default along with permissive CORS ( allow\\_origins=\\[\u201c\\*\u201d\\] ) meaning the following API endpoint is exposed to every network interface the daemon is connected to.\n\n# PoC\n\n1. Start the daemon in simulation mode (command depends on the installed environment):\n\n```shell\n .venv/bin/mjpython -m reachy_mini.daemon.app.main --sim --no-media\n```\n\n2. After that check that the media upload API endpoint is activated and you can upload a wav file:\n\n```shell\ncurl -X POST http://\u003cdaemon_domain\u003e:\u003cdaemon_port\u003e/api/media/sounds/upload \\\n -F \"file=@/path/to/your/file.wav\"\n```\n\n3. Now attempt to create a \u201c.sh\u201d file containing a script and upload it:\n\n```shell\ncurl -X POST http://\u003cdaemon_domain\u003e:\u003cdaemon_port\u003e/api/media/sounds/upload \\\n -F \"file=@/path/to/your/script.sh\"\n```\n\n4. Now error message will be received and you will see that the script file was successfully uploaded to disk.\n\n# Impact\n\nDue to this issue, an attacker can upload malicious files instead of the intended sounds files, harming the integrity of the stored data and allowing an attacker to propagate a foothold in cases another vulnerabilities would arise.\n\n## Fix suggestion\n\nPerform the following check on the API endpoint:\n\n1. Validate that the file extension contains only desired extensions (allow-list approach). \n2. Validate that the uploaded file\u2019s content matches the desired extension (Magic numbers, and known file structure per file type). \n3. Enforce authentication on the file upload endpoint.\n\n## Credit\n\nThe vulnerability was discovered by Natan Nehorai of the JFrog Vulnerability Research team.",
"id": "PYSEC-2026-3916",
"modified": "2026-09-10T11:02:31.717357Z",
"published": "2026-09-10T09:44:56.500259Z",
"references": [
{
"type": "WEB",
"url": "https://github.com/pollen-robotics/reachy_mini/security/advisories/GHSA-m2pc-3q4q-w6jr"
},
{
"type": "WEB",
"url": "https://github.com/pollen-robotics/reachy_mini/commit/984c7723b3ec5da63f4e0a2bcf9f120ceb563e04"
},
{
"type": "PACKAGE",
"url": "https://github.com/pollen-robotics/reachy_mini"
},
{
"type": "PACKAGE",
"url": "https://pypi.org/project/reachy-mini"
},
{
"type": "ADVISORY",
"url": "https://github.com/advisories/GHSA-m2pc-3q4q-w6jr"
},
{
"type": "ADVISORY",
"url": "https://nvd.nist.gov/vuln/detail/CVE-2026-55419"
}
],
"severity": [
{
"score": "CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:L/A:N",
"type": "CVSS_V3"
}
],
"summary": "reachy_mini Allows Unrestricted Upload of File with Dangerous Type"
}
Drag nodes to arrange them. Double-click a related vulnerability to open it, or a source to visit its link. The Physics button on the left tunes the layout and the simulation; the legend at the bottom left hides or shows nodes by kind and by sighting type.
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.
Vulnerability timeline
Sightings and recorded dates, grouped by day · UTC
How durations are calculated
Following the VTEM framework (Black Hat USA 2015), time to patch starts when the vendor or a coordinator was informed, total time to patch at discovery, and disclosure to patch at public disclosure; vendor response time runs from the report to its acknowledgement. Time to exploit and time unpatched start at the earliest discovery or report, falling back to public disclosure. Unpatched time ends at the first recorded patch, or the calculation time when no patch is recorded. Exploitation to patch shows the interval between those two events; a published proof of concept is a public exploit, not exploitation.
Each result shows its source dates. A negative interval means the end event came first; when date-only inputs leave the order open, the result says so. Date-only inputs are approximate. The earliest evidence wins: KEV report dates and sightings are labelled as such. Records describing another vulnerability are listed but never calculated from, and a PoC alone does not establish exploitation. Missing patch evidence does not prove that no patch exists. Filters below change the event list only.
Loading timeline…
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.