MCP Run Python Deno Sandbox Misconfiguration Allows SSRF Attacks via Localhost Access

Description

Impact

Server-Side Request Forgery (SSRF):
A security vulnerability exists in the mcp-run-python tool (specifically within the Pydantic-AI integration) due to an overly permissive Deno sandbox configuration.

The tool configures the Deno runtime—which is intended to isolate the execution of untrusted Python code—with network permissions that include access to the host's loopback interface (localhost). Consequently, malicious Python code executed through the tool can bypass network isolation and send HTTP requests to internal services running on the host machine. This allows attackers to interact with local APIs, databases, or cloud metadata services that should not be accessible from the sandbox.

Patches

No Patch Available:
The mcp-run-python project is currently archived and maintainers have indicated it is unlikely to receive a fix.

Recommendation:
Users are strongly advised to immediately stop using this package.
If functionality is required, users must migrate to an alternative execution environment that enforces strict network isolation (e.g., blocking all outbound traffic or explicitly denying access to 127.0.0.1/::1).

Workarounds

There are no configuration-based workarounds provided by the package itself. Remediation requires modifying the source code to restrict the Deno permissions (specifically removing or narrowing the --allow-net flag) or moving the execution to a container with no network access.

Resources

Basic information

Type
reviewed
Severity
medium
Advisory on GitHub
Open advisory ↗
Repository advisory
Source code
Browse source ↗
Published (advisory)
2026-02-09 09:30:22 UTC
Updated
2026-02-11 19:47:54 UTC
GitHub reviewed
2026-02-11 19:47:51 UTC
NVD published
2026-02-09

EPSS Score

Score Percentile
0.01% 2.04%

CVSS Scores

Base score Version Severity Vector
5.8 3.1
CVSS:3.1/AV:N/AC:H/PR:N/UI:R/S:C/C:L/I:L/A:L Click to expand
Attack vector (AV:N)
Could be attacked over the internet or any normal routed network—not just someone sitting at the machine.
Attack complexity (AC:H)
Even with access, the exploit needs extra luck, timing, or a fussy environment to actually work.
Privileges required (PR:N)
No account or special rights needed—anonymous or random user is enough.
User interaction (UI:R)
A real person has to do something—click, install, enable—otherwise it doesn’t land.
Scope (S:C)
Breaking this can reach past the original component and bite other resources—bigger blast radius.
Confidentiality (C:L)
Some sensitive info could get out, but not a total data dump.
Integrity (I:L)
Attackers could change some data, but it’s limited—not everything goes.
Availability (A:L)
Might cause slowdowns, glitches, or partial disruption—not a full brick.

Identifiers

CWEs

CWE id Name
CWE-918 Server-Side Request Forgery (SSRF)

Credits

  • saivarun3407 (analyst)

Affected packages (1)

Vulnerable version ranges and first patched releases as published by GitHub.

Ecosystem Package Vulnerable range First patched Vulnerable functions
pip mcp-run-python <= 0.0.22

References

cvelogic Threat Intelligence