AI agents invoke milvus-release-collection to trigger actions in Milvus MCP Server. 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.
Releasing a collection from memory is an operational action that affects the runtime state of the database — it makes the collection unavailable for queries until reloaded. This is not a simple read or write; it triggers an external operation with significant impact on service availability. It is not irreversible (the collection data is not deleted, just unloaded), so it falls under Execute rather than Destructive.
From the tool's definition Release a collection from memory
Documented attack patterns abuse exactly the kind of access milvus-release-collection gives an agent:
PolicyLayer is an MCP gateway — it sits between your AI agents and Milvus MCP Server, and nothing reaches the server without passing your rules. This is the rule we recommend for milvus-release-collection:
{
"version": "1",
"default": "deny",
"tools": {
"milvus-release-collection": {
"limits": [
{
"counter": "milvus-release-collection_rate",
"window": "minute",
"max": 10,
"scope": "grant"
}
]
}
}
} milvus-release-collection 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.
Release a collection from memory. It is categorised as a Execute tool in the Milvus MCP Server MCP Server, which means it can trigger actions or run processes. Use rate limits and argument validation.
Register the Milvus MCP Server MCP server in PolicyLayer and add a rule for milvus-release-collection: 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 Milvus MCP Server. Nothing to install.
milvus-release-collection 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 milvus-release-collection 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 milvus-release-collection. 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.
milvus-release-collection is provided by the Milvus MCP Server MCP server (stephen37/mcp-server-milvus). PolicyLayer sits as a proxy in front of this server to enforce policies before tool calls reach the server.
Start from Milvus MCP Server, 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.
21 Milvus MCP Server tools catalogued and risk-classified — across an index of 43,000+ MCP servers.