This record as markdown: /tools/io-github-portel-dev-ncp/inference.md
What inference does on Ncp
AI agents invoke inference to trigger actions in Ncp. 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.
Why inference is rated High
The tool invokes external ML model inference, which qualifies as Execute (code/operation execution) rather than Read because inference can have side effects depending on the model and input arguments.
From the tool's definition Tool performs 'Run inference on Hugging Face model' — this executes machine learning operations whose effects depend on the input model and data provided, similar to executing arbitrary code or scripts.
Attacks that exploit this kind of access
The rule that runs inference safely
PolicyLayer is an MCP gateway: it sits between your AI agents and Ncp, and checks every tool call against a rule you set before the call runs. Nothing changes on the server itself. For inference, this is the rule to start with:
inference 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.
The button opens the PolicyLayer dashboard: create your workspace, connect Ncp, apply this rule, and every inference call is checked against it from then on.
Questions about inference
Run inference on Hugging Face model. It is categorised as a Execute tool in the Ncp MCP Server, which means it can trigger actions or run processes. Use rate limits and argument validation.
Register the Ncp MCP server in PolicyLayer and add a rule for inference: 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 Ncp. Nothing to install.
inference 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 inference 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 inference. 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.
inference is provided by the Ncp MCP server (@portel/ncp). PolicyLayer sits as a proxy in front of this server to enforce policies before tool calls reach the server.
More on Ncp, and thousands of servers like it.
This server
Across the catalogue