AI agents invoke swap_model to trigger actions in Mcp Llama Swap. What it does depends on the arguments the agent supplies, and its effects often reach beyond the immediate call — builds kicked off, notifications sent, workflows started.
This tool executes a system-level operation (via launchctl or systemd) to swap the running llama.cpp model. It's not merely writing configuration data — it actively restarts or reconfigures a system service, which is an external execution with significant blast radius: a malicious or incorrect swap could redirect all AI inference to an untrusted or compromised model, affecting the integrity of the entire session.
From the tool's definition 'Swap to a different llama.cpp model' — triggers an external operation (hot-swapping a running model via launchctl or systemd) that changes the active inference backend in a live session.
Documented attack patterns abuse exactly the kind of access swap_model gives an agent:
PolicyLayer is an MCP gateway — it sits between your AI agents and Mcp Llama Swap, and nothing reaches the server without passing your rules. This is the rule we recommend for swap_model:
{
"version": "1",
"default": "deny",
"tools": {
"swap_model": {
"limits": [
{
"counter": "swap_model_rate",
"window": "minute",
"max": 10,
"scope": "grant"
}
]
}
}
}swap_model stays usable, but rate-capped — a runaway agent can't fire it dozens of times a minute. Everything else on the server is denied unless you say otherwise.
Free to start. No card required.
Swap to a different llama.cpp model. It is categorised as a Execute tool in the Mcp Llama Swap MCP Server, which means it can trigger actions or run processes. Use rate limits and argument validation.
Register the Mcp Llama Swap MCP server in PolicyLayer and add a rule for swap_model: allow, deny, rate-limit, or require approval. Point your MCP client at the PolicyLayer proxy URL and the rule is enforced on every call, before it reaches Mcp Llama Swap. Nothing to install.
swap_model is a Execute tool with high risk. Execute tools should be rate-limited and have argument validation enabled.
Yes. Add a rate_limit block to the swap_model rule in your PolicyLayer policy. For example, setting max: 10 and window: 60 limits the tool to 10 calls per minute. Rate limits are tracked per agent session and reset automatically.
Set action: deny in the PolicyLayer policy for swap_model. The AI agent will receive a policy violation error and cannot call the tool. You can also include a reason field to explain why the tool is blocked.
swap_model is provided by the Mcp Llama Swap MCP server (oussama-kh/mcp-llama-swap). PolicyLayer sits as a proxy in front of this server to enforce policies before tool calls reach the server.
Start from Mcp Llama Swap, add the rest of your stack, and see everything your agents can call. Then put policy on all of it.
Free to start. No card required.
4 Mcp Llama Swap tools catalogued and risk-classified — across an index of 43,000+ MCP servers.