atomic-agents-stack: HTTP MCP catalog accepts cleartext http and spawns catalog-supplied commands (MITM to RCE)
- Product
- atomic-agents-stack
- Affected
- atomic-agents-stack <= 1.0.0
- Fixed in
- 1.1.0
Last verified against sources: October 26, 2026
Am I affected?
atomic-agents-stack: affected atomic-agents-stack <= 1.0.0. Fixed in 1.1.0.
- Check your version: run
pip show atomic-agents-stack. Versions <= 1.0.0 of atomic-agents-stack are affected. [1]
What to do now
- Upgrade to the fixed version: run
pip install --upgrade "atomic-agents-stack>=1.1.0"(first patched version 1.1.0). [1] - Require
httpsby default and gatehttp://behind a loud explicit opt-in. [1] - Allowlist the resolved command basename (or require confirmation) before any registry-sourced subprocess spawn. [1]
General practice, not from the source: if the affected software could read tokens or credentials, rotate them after upgrading; pin versions and review changes before updating tools your agents run.
What happened
- The HTTP MCP server-registry backend factory (
atomic_agents/mcp_registry/http.py,make_http_mcp_server_registry_backend_from_url) accepts bothhttpandhttpsschemes. [1] - Catalog entries carry
command/argsthat are type-validated but content-unrestricted, and are later spawned as local stdio subprocesses byMCPClientPool. [1] - The Policy MCP allowlist is not a default mitigation (
mcp_allow_fndefaults to None), so absent an operator-authored allowlist every resolved spec connects. [1]
Impact
- Over a cleartext
http://catalog URL, a network man-in-the-middle can rewrite the catalog response to inject an arbitrarycommand/argsand obtain code execution on the agent host, with no LLM involvement. [1]
What was done
- Require
httpsby default and gatehttp://behind a loud explicit opt-in. [1] - Defense-in-depth: allowlist the resolved command basename (or require confirmation) before any registry-sourced subprocess spawn. [1]
- Document the consequence in spec/36. [1]
- pip/atomic-agents-stack affected <= 1.0.0 patched 1.1.0. [1]
Sources
- GHSA-xhcr-cqfr-m3hv — GitHub Advisory Database
Every statement above was checked against these sources before publishing. Spotted an error? Tell us and we will correct it with a dated note.
Share advisory brief with your team
Copy a plain-text summary for Slack, GitHub issues, or team security channels.
atomic-agents-stack: HTTP MCP catalog accepts cleartext http and spawns catalog-supplied commands (MITM to RCE)
Product: atomic-agents-stack | Affected: atomic-agents-stack <= 1.0.0 | Fixed in: 1.1.0
Summary: The HTTP MCP server-registry backend factory (`atomic_agents/mcp_registry/http.py`, `make_http_mcp_server_registry_backend_from_url`) accepts both `http` and `https` schemes.
Source & remediation on AgentArmorHQ: https://agentarmorhq.com/incidents/2026/08/ghsa-xhcr-cqfr-m3hv-atomic-agents-stackCorrections
Report an error or request a correction. We review requests within 5 business days and add a dated correction note.
Discussions for atomic-agents-stack
Ask questions or share configuration tips. Governed by our Community Rules & Terms.
No posts match this filter yet.
Be the first to ask a question or share a working tip.