Zyrabit Platform provides sovereign AI infrastructure for enterprise retrieval-augmented generation.
CVE-2026-130154
Zyrabit Platform provides sovereign AI infrastructure for enterprise retrieval-augmented generation. Prior to 2.4.4, importtovault(sourcepath, destinationname) accepts both caller-controlled paths, but the content-pattern guard does not confine destinationname. The vulnerable tool is implemented in zyrabit-slm/api-rag/app/domain/services/mcpservice.py, where Path(DOCSDIR) / destinationname is passed to shutil.copy2, allowing traversal or an absolute destination to write outside the vault. An unrestricted sourcepath can be copied into the vault and retrieved through securequery, disclosing files readable by the service; files that do not match the executable-pattern filter pass that filter. When ENABLELEGACYEXTENSIONS=true, zyrabit-slm/api-rag/app/main.py registers POST /mcp/rpc without getcurrentuser, allowing unauthenticated network tool calls; this optional exposure is disabled by default and is not required for local invocation. If an attacker can stage chosen source content and write to an execution-triggered destination, the advisory describes conditional process execution. This issue is fixed in version 2.4.4.
Assessment
- CVSS
- 9.1 (v3.1)
- Severity
- Critical
- Known exploited
- Not listed in CISA KEV
- Exploit availability
- Nothing found in Exploit-DB, Metasploit, Nuclei or public PoC repositories
- Published
- 18 November 2026
- Last updated
- 18 November 2026
- Affected
- Zyrabit-tech · zyrabit-SLM
- Assigner
- GitHub_M
CVSS vectorCVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:N
At the source
References
Community proofs of concept
No community proof of concept yet. If you have one for a published vulnerability and the right to share it, add it below — it will appear here once reviewed.
Submit a proof of concept
Only for publicly disclosed vulnerabilities, and only material from testing you were authorised to perform. This platform does not buy, sell, or broker exploits — see earning from vulnerability research for legitimate bounty programs. Submissions are reviewed before they appear. Text only; do not include live credentials, personal data, or anything targeting systems you are not permitted to test.
Query this programmatically
Agents can read this through the MCP endpoint, or over HTTP:
curl https://cve.zerocrew.in/api/cve/CVE-2026-130154