{"id":"CVE-2026-91988","aliases":["EUVD-2026-78659","GHSA-wr36-693x-8mgc","GHSA-xhcr-cqfr-m3hv"],"status":"published","title":"atomic-agents-stack before 1.1.0 Remote Code Execution via HTTP MCP","description":"atomic-agents-stack before 1.1.0 accepts cleartext HTTP schemes in the HTTP MCP server-registry backend factory, allowing network man-in-the-middle attackers to rewrite catalog responses. Attackers can inject arbitrary command and argument values that are spawned as local subprocesses by MCPClientPool to achieve code execution on the agent host.","severity":{"version":"4.0","score":9.2,"vector":"CVSS:4.0/AV:N/AC:H/AT:P/PR:N/UI:N/VC:H/VI:H/VA:H/SC:N/SI:N/SA:N","level":"critical","assessor":"cna","from":"cve"},"epss":{"day":"2026-11-12","score":0.00318,"percentile":0.25124,"model":"v2026.06.15"},"facts":{"in_kev":false,"known_exploited":false,"public_exploit":false,"fix_available":true},"cwes":["CWE-319","CWE-494"],"published":"2026-08-17T21:49:55Z","modified":"2026-10-28T02:30:41Z","fixes":[{"ecosystem":"PyPI","package":"atomic-agents-stack","fixed_in":["1.1.0"],"sources":["ghsa","osv"]}],"affected":[{"source":"ghsa","ecosystem":"PyPI","package":"atomic-agents-stack","ranges":[{"type":"ECOSYSTEM","events":[{"introduced":"0"},{"last_affected":"1.0.0"},{"fixed":"1.1.0"}]}]},{"source":"osv","ecosystem":"PyPI","package":"atomic-agents-stack","ranges":[{"type":"ECOSYSTEM","events":[{"introduced":"0"},{"fixed":"1.1.0"}]}],"versions":["1.0.0"]}],"products":[{"vendor":"dep0we","product":"atomic-agents-stack"}],"references":[{"url":"https://github.com/dep0we/atomic-agents-stack/security/advisories/GHSA-xhcr-cqfr-m3hv","tags":["vendor-advisory","advisory","web"]},{"url":"https://www.vulncheck.com/advisories/atomic-agents-stack-before-1.1.0-remote-code-execution-via-http-mcp","tags":["third-party-advisory","advisory"]},{"url":"https://nvd.nist.gov/vuln/detail/CVE-2026-91988","tags":["advisory"]},{"url":"https://github.com/advisories/GHSA-wr36-693x-8mgc"},{"url":"https://github.com/dep0we/atomic-agents-stack/releases#release-v1.1.0","tags":["web"]},{"url":"https://github.com/advisories/GHSA-xhcr-cqfr-m3hv"},{"url":"https://github.com/CVEProject/cvelistV5/tree/main/cves/2026/91xxx/CVE-2026-91988.json","tags":["advisory"]},{"url":"https://github.com/dep0we/atomic-agents-stack","tags":["package"]}],"sources":{"cve":[{"id":"CVE-2026-91988","title":"atomic-agents-stack before 1.1.0 Remote Code Execution via HTTP MCP","description":"atomic-agents-stack before 1.1.0 accepts cleartext HTTP schemes in the HTTP MCP server-registry backend factory, allowing network man-in-the-middle attackers to rewrite catalog responses. Attackers can inject arbitrary command and argument values that are spawned as local subprocesses by MCPClientPool to achieve code execution on the agent host.","status":"published","published":"2026-09-15T15:18:30.347Z","modified":"2026-09-15T15:50:03.261Z","cvss":[{"version":"4.0","score":9.2,"vector":"CVSS:4.0/AV:N/AC:H/AT:P/PR:N/UI:N/VC:H/VI:H/VA:H/SC:N/SI:N/SA:N","level":"critical","assessor":"cna"},{"version":"3.1","score":8.1,"vector":"CVSS:3.1/AV:N/AC:H/PR:N/UI:N/S:U/C:H/I:H/A:H","level":"high","assessor":"cna"}],"cwes":["CWE-319"],"updated_at":"2026-11-02T14:19:53.115938Z"}],"euvd":[{"id":"EUVD-2026-78659","description":"atomic-agents-stack before 1.1.0 accepts cleartext HTTP schemes in the HTTP MCP server-registry backend factory, allowing network man-in-the-middle attackers to rewrite catalog responses. Attackers can inject arbitrary command and argument values that are spawned as local subprocesses by MCPClientPool to achieve code execution on the agent host.","status":"published","published":"2026-09-15T15:18:30Z","modified":"2026-09-15T15:50:03Z","cvss":[{"version":"4.0","score":9.2,"vector":"CVSS:4.0/AV:N/AC:H/AT:P/PR:N/UI:N/VC:H/VI:H/VA:H/SC:N/SI:N/SA:N","level":"critical"}],"updated_at":"2026-11-02T17:28:17.768191Z"}],"ghsa":[{"id":"GHSA-wr36-693x-8mgc","title":"atomic-agents-stack before 1.1.0 accepts cleartext HTTP schemes in the HTTP MCP server-registry...","description":"atomic-agents-stack before 1.1.0 accepts cleartext HTTP schemes in the HTTP MCP server-registry backend factory, allowing network man-in-the-middle attackers to rewrite catalog responses. Attackers can inject arbitrary command and argument values that are spawned as local subprocesses by MCPClientPool to achieve code execution on the agent host.","status":"published","published":"2026-09-15T18:32:31Z","modified":"2026-09-15T18:32:42Z","cvss":[{"version":"4.0","score":9.2,"vector":"CVSS:4.0/AV:N/AC:H/AT:P/PR:N/UI:N/VC:H/VI:H/VA:H/SC:N/SI:N/SA:N/E:X/CR:X/IR:X/AR:X/MAV:X/MAC:X/MAT:X/MPR:X/MUI:X/MVC:X/MVI:X/MVA:X/MSC:X/MSI:X/MSA:X/S:X/AU:X/R:X/V:X/RE:X/U:X","level":"critical"},{"version":"3.1","score":8.1,"vector":"CVSS:3.1/AV:N/AC:H/PR:N/UI:N/S:U/C:H/I:H/A:H","level":"high"}],"cwes":["CWE-319"],"updated_at":"2026-11-02T15:11:54.818367Z"},{"id":"GHSA-xhcr-cqfr-m3hv","title":"atomic-agents-stack: HTTP MCP catalog accepts cleartext http and spawns catalog-supplied commands (MITM to RCE)","description":"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. Catalog entries carry `command`/`args` that are type-validated but content-unrestricted, and are later spawned as local stdio subprocesses by `MCPClientPool`. Over a cleartext `http://` catalog URL, a network man-in-the-middle can rewrite the catalog response to inject an arbitrary `command`/`args` and obtain code execution on the agent host, with no LLM involvement. The Policy MCP allowlist is not a default mitigation (`mcp_allow_fn` defaults to None), so absent an operator-authored allowlist every resolved spec connects.\n\n**Affected:** `mcp_registry/http.py`, all versions through 1.0.0. (The `https` path is sound: `httpx` defaults to `verify=True`, `follow_redirects=False`.)\n\n**Fix:** require `https` by default and gate `http://` behind a loud explicit opt-in. Defense-in-depth: allowlist the resolved command basename (or require confirmation) before any registry-sourced subprocess spawn. Document the consequence in spec/36.","status":"published","published":"2026-08-17T21:49:55Z","modified":"2026-08-17T21:49:55Z","cvss":[{"version":"4.0","score":8.7,"vector":"CVSS:4.0/AV:N/AC:L/AT:N/PR:N/UI:N/VC:N/VI:H/VA:N/SC:N/SI:N/SA:N","level":"high"}],"cwes":["CWE-319","CWE-494"],"updated_at":"2026-11-02T14:41:01.23108Z"}],"nvd":[{"id":"CVE-2026-91988","description":"atomic-agents-stack before 1.1.0 accepts cleartext HTTP schemes in the HTTP MCP server-registry backend factory, allowing network man-in-the-middle attackers to rewrite catalog responses. Attackers can inject arbitrary command and argument values that are spawned as local subprocesses by MCPClientPool to achieve code execution on the agent host.","status":"published","published":"2026-09-15T16:17:57.45Z","modified":"2026-09-24T20:43:32.537Z","cvss":[{"version":"4.0","score":9.2,"vector":"CVSS:4.0/AV:N/AC:H/AT:P/PR:N/UI:N/VC:H/VI:H/VA:H/SC:N/SI:N/SA:N/E:X/CR:X/IR:X/AR:X/MAV:X/MAC:X/MAT:X/MPR:X/MUI:X/MVC:X/MVI:X/MVA:X/MSC:X/MSI:X/MSA:X/S:X/AU:X/R:X/V:X/RE:X/U:X","level":"critical","assessor":"secondary"},{"version":"3.1","score":8.1,"vector":"CVSS:3.1/AV:N/AC:H/PR:N/UI:N/S:U/C:H/I:H/A:H","level":"high","assessor":"primary"}],"cwes":["CWE-319"],"updated_at":"2026-11-02T14:12:47.668237Z"}],"osv":[{"id":"CVE-2026-91988","title":"atomic-agents-stack before 1.1.0 Remote Code Execution via HTTP MCP","description":"atomic-agents-stack before 1.1.0 accepts cleartext HTTP schemes in the HTTP MCP server-registry backend factory, allowing network man-in-the-middle attackers to rewrite catalog responses. Attackers can inject arbitrary command and argument values that are spawned as local subprocesses by MCPClientPool to achieve code execution on the agent host.","status":"published","published":"2026-09-15T15:18:30.347Z","modified":"2026-10-28T02:30:41.432076971Z","cvss":[{"version":"4.0","score":9.2,"vector":"CVSS:4.0/AV:N/AC:H/AT:P/PR:N/UI:N/VC:H/VI:H/VA:H/SC:N/SI:N/SA:N","level":"critical"}],"cwes":["CWE-319"],"updated_at":"2026-11-02T15:45:20.269367Z"},{"id":"GHSA-xhcr-cqfr-m3hv","title":"atomic-agents-stack: HTTP MCP catalog accepts cleartext http and spawns catalog-supplied commands (MITM to RCE)","description":"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. Catalog entries carry `command`/`args` that are type-validated but content-unrestricted, and are later spawned as local stdio subprocesses by `MCPClientPool`. Over a cleartext `http://` catalog URL, a network man-in-the-middle can rewrite the catalog response to inject an arbitrary `command`/`args` and obtain code execution on the agent host, with no LLM involvement. The Policy MCP allowlist is not a default mitigation (`mcp_allow_fn` defaults to None), so absent an operator-authored allowlist every resolved spec connects.\n\n**Affected:** `mcp_registry/http.py`, all versions through 1.0.0. (The `https` path is sound: `httpx` defaults to `verify=True`, `follow_redirects=False`.)\n\n**Fix:** require `https` by default and gate `http://` behind a loud explicit opt-in. Defense-in-depth: allowlist the resolved command basename (or require confirmation) before any registry-sourced subprocess spawn. Document the consequence in spec/36.","status":"published","published":"2026-08-17T21:49:55Z","modified":"2026-09-16T03:56:06.204188287Z","cvss":[{"version":"4.0","score":8.7,"vector":"CVSS:4.0/AV:N/AC:L/AT:N/PR:N/UI:N/VC:N/VI:H/VA:N/SC:N/SI:N/SA:N","level":"high"}],"cwes":["CWE-319","CWE-494"],"updated_at":"2026-11-02T15:46:59.835583Z"}]}}