AI agents invoke start_neb to trigger actions in Fairchem. 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.
NEB is a computationally expensive simulation that runs external calculations and modifies system state (calculates energy barriers, reaction pathways). While not permanently destructive to stored data, it executes arbitrary computational operations whose outcomes depend entirely on user-supplied arguments (atomic structures, force fields via sibling tools like 'attach_emt' or 'attach_lammps').
From the tool's definition Tool is part of a computational simulation server that executes NEB (Nudged Elastic Band) calculations. Context clues: (1) sibling tool 'execute' suggests this server runs code/simulations; (2) 'start_neb' initiates a complex multi-step external operation…
Attacks that exploit this kind of access
start_neb. It is categorised as a Execute tool in the Fairchem MCP Server, which means it can trigger actions or run processes. Use rate limits and argument validation.
Register the Fairchem MCP server in PolicyLayer and add a rule for start_neb: 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 Fairchem. Nothing to install.
start_neb 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 start_neb 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 start_neb. 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.
start_neb is provided by the Fairchem MCP server (jkitchin/fairchem-mcp). PolicyLayer sits as a proxy in front of this server to enforce policies before tool calls reach the server.
Every MCP server has a record like this.
Type a name, get the same breakdown: verified identity, auth posture, risk grade, capabilities, recommended policy.
Teams ship this data inside their own products. See what a licence covers →