GHSA-9VHV-G5HF-M7M7
Vulnerability from github – Published: 2026-11-09 22:01 – Updated: 2026-11-09 22:01Summary
fast-agent serve exposes the agent over MCP. By default it uses HTTP transport bound to 0.0.0.0 (all interfaces) on port 8000 with no authentication (auth=None) and prints no warning. Any host that can reach the port can call the exposed agent tool. When the server is started with --shell (which is -x, the flag used in the README quickstart uvx fast-agent mcp@latest -x), the agent has a local execute shell tool, so an unauthenticated remote client can drive the agent into running arbitrary OS commands — unauthenticated RCE.
Affected code
cli/commands/serve.py— transport defaults to HTTP;--hostdefaults to0.0.0.0; port 8000.mcp/server/agent_server.py—auth_provider = NoneunlessFAST_AGENT_SERVE_OAUTH=huggingface;run_async(transport="http", host=host, ...)withhostdefaulting to0.0.0.0.tools/local_shell_executor.py— the shell tool runs commands viacreate_subprocess_shell.
Verified (live, faithful — published 0.7.20 in Docker python:3.13-slim, stock unauthenticated client)
1) Unauthenticated exposure (core defect): a client with no Authorization header completed
initialize and tools/list (-> ['agent']) against http://HOST:8000/mcp. Model-independent.
2) Unauthenticated RCE with --shell, demonstrated two ways:
- Real production model (gpt-4o-mini): a single plain-language message asking the agent to run
a command via its shell tool -> the model autonomously called execute and the command output
(process identity showing it ran as root, the hostname, and the first lines of a sensitive
system file) was returned to the unauthenticated caller.
- Key-free deterministic (passthrough model): the tool-call directive invoked the shell tool
directly, same result.
The server-side permission prompt did not block this in non-interactive serve mode.
Proof of concept
Environment: Docker python:3.13-slim, pip install fast-agent-mcp -> 0.7.20.
# server (real model + the documented shell flag; 0.0.0.0 is also the built-in default)
OPENAI_API_KEY=*** fast-agent serve --model gpt-4o-mini --shell --host 0.0.0.0 --port 8000
# -> "Uvicorn running on http://0.0.0.0:8000" (no auth, no warning)
# attacker: unauthenticated MCP HTTP client (NO Authorization header) to http://HOST:8000/mcp
# initialize() -> 200 OK (server fast-agent-MCP-Server 0.7.20)
# tools/list -> ['agent']
# tools/call name="agent" arguments={"message":
# "Using your execute/shell tool, run exactly this command and return its raw stdout: "
# "id; echo PWNED=$(hostname); cat /etc/passwd | head -2"}
#
# response returned to the UNAUTHENTICATED caller:
# uid=0(root) gid=0(root) groups=0(root)
# PWNED=<host>
# <first lines of /etc/passwd>
Also reproduced key-free with --model passthrough and the message ***CALL_TOOL execute {"command":"id"} (passthrough invokes the tool deterministically). Full transcript + client scripts were sent by email.
Impact
An unauthenticated client that can reach the serve port controls the agent: with --shell this is remote command execution as the server user; without --shell it is unauthenticated use of the operator's LLM and any configured MCP servers' tools. The 0.0.0.0 default makes the server reachable beyond localhost as soon as it runs on a non-isolated host. Preconditions (serve + shell) are noted above and are why I defer the exact rating to you.
Remediation
- Default the bind host to
127.0.0.1; require an explicit--host 0.0.0.0opt-in for non-loopback. - Print a loud warning when binding a non-loopback interface without authentication.
{
"affected": [
{
"database_specific": {
"last_known_affected_version_range": "\u003c= 0.7.20"
},
"package": {
"ecosystem": "PyPI",
"name": "fast-agent-mcp"
},
"ranges": [
{
"events": [
{
"introduced": "0"
},
{
"fixed": "0.7.21"
}
],
"type": "ECOSYSTEM"
}
]
}
],
"aliases": [
"CVE-2026-102610"
],
"database_specific": {
"cwe_ids": [
"CWE-306",
"CWE-78"
],
"github_reviewed": true,
"github_reviewed_at": "2026-11-09T22:01:34Z",
"nvd_published_at": "2026-11-06T21:17:05Z",
"severity": "MODERATE"
},
"details": "## Summary\n`fast-agent serve` exposes the agent over MCP. By default it uses HTTP transport bound to `0.0.0.0` (all interfaces) on port 8000 with **no authentication** (`auth=None`) and prints no warning. Any host that can reach the port can call the exposed `agent` tool. When the server is started with `--shell` (which is `-x`, the flag used in the README quickstart `uvx fast-agent mcp@latest -x`), the agent has a local `execute` shell tool, so an unauthenticated remote client can drive the agent into running arbitrary OS commands \u2014 unauthenticated RCE.\n\n## Affected code\n- `cli/commands/serve.py` \u2014 transport defaults to HTTP; `--host` defaults to `0.0.0.0`; port 8000.\n- `mcp/server/agent_server.py` \u2014 `auth_provider = None` unless `FAST_AGENT_SERVE_OAUTH=huggingface`;\n `run_async(transport=\"http\", host=host, ...)` with `host` defaulting to `0.0.0.0`.\n- `tools/local_shell_executor.py` \u2014 the shell tool runs commands via `create_subprocess_shell`.\n\n## Verified (live, faithful \u2014 published 0.7.20 in Docker python:3.13-slim, stock unauthenticated client)\n1) Unauthenticated exposure (core defect): a client with no Authorization header completed\n `initialize` and `tools/list` (-\u003e `[\u0027agent\u0027]`) against `http://HOST:8000/mcp`. Model-independent.\n2) Unauthenticated RCE with `--shell`, demonstrated two ways:\n - Real production model (`gpt-4o-mini`): a single plain-language message asking the agent to run\n a command via its shell tool -\u003e the model autonomously called `execute` and the command output\n (process identity showing it ran as root, the hostname, and the first lines of a sensitive\n system file) was returned to the unauthenticated caller.\n - Key-free deterministic (`passthrough` model): the tool-call directive invoked the shell tool\n directly, same result.\n The server-side permission prompt did not block this in non-interactive serve mode.\n\n## Proof of concept\nEnvironment: Docker `python:3.13-slim`, `pip install fast-agent-mcp` -\u003e 0.7.20.\n\n```\n# server (real model + the documented shell flag; 0.0.0.0 is also the built-in default)\nOPENAI_API_KEY=*** fast-agent serve --model gpt-4o-mini --shell --host 0.0.0.0 --port 8000\n# -\u003e \"Uvicorn running on http://0.0.0.0:8000\" (no auth, no warning)\n\n# attacker: unauthenticated MCP HTTP client (NO Authorization header) to http://HOST:8000/mcp\n# initialize() -\u003e 200 OK (server fast-agent-MCP-Server 0.7.20)\n# tools/list -\u003e [\u0027agent\u0027]\n# tools/call name=\"agent\" arguments={\"message\":\n# \"Using your execute/shell tool, run exactly this command and return its raw stdout: \"\n# \"id; echo PWNED=$(hostname); cat /etc/passwd | head -2\"}\n#\n# response returned to the UNAUTHENTICATED caller:\n# uid=0(root) gid=0(root) groups=0(root)\n# PWNED=\u003chost\u003e\n# \u003cfirst lines of /etc/passwd\u003e\n```\nAlso reproduced key-free with `--model passthrough` and the message `***CALL_TOOL execute {\"command\":\"id\"}` (passthrough invokes the tool deterministically). Full transcript + client scripts were sent by email.\n\n## Impact\nAn unauthenticated client that can reach the serve port controls the agent: with `--shell` this is remote command execution as the server user; without `--shell` it is unauthenticated use of the operator\u0027s LLM and any configured MCP servers\u0027 tools. The `0.0.0.0` default makes the server reachable beyond localhost as soon as it runs on a non-isolated host. Preconditions (serve + shell) are noted above and are why I defer the exact rating to you.\n\n## Remediation\n1. Default the bind host to `127.0.0.1`; require an explicit `--host 0.0.0.0` opt-in for non-loopback.\n2. Print a loud warning when binding a non-loopback interface without authentication.",
"id": "GHSA-9vhv-g5hf-m7m7",
"modified": "2026-11-09T22:01:34Z",
"published": "2026-11-09T22:01:34Z",
"references": [
{
"type": "WEB",
"url": "https://github.com/evalstate/fast-agent/security/advisories/GHSA-9vhv-g5hf-m7m7"
},
{
"type": "ADVISORY",
"url": "https://nvd.nist.gov/vuln/detail/CVE-2026-102610"
},
{
"type": "WEB",
"url": "https://github.com/evalstate/fast-agent/pull/828"
},
{
"type": "WEB",
"url": "https://github.com/evalstate/fast-agent/commit/ba99f89c3ec7f1ef1c8b1ebce381c12b60dfea29"
},
{
"type": "PACKAGE",
"url": "https://github.com/evalstate/fast-agent"
},
{
"type": "WEB",
"url": "https://github.com/evalstate/fast-agent/releases/tag/v0.7.21"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:3.1/AV:N/AC:H/PR:H/UI:R/S:U/C:H/I:H/A:H",
"type": "CVSS_V3"
}
],
"summary": "fast-agent: `fast-agent serve` binds HTTP to 0.0.0.0 with no authentication by default; with --shell this is unauthenticated remote command execution"
}
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.