AI agents call get_run_details to retrieve information from Mcp Wandb without modifying anything — typically the context-gathering step in research, monitoring, and reporting workflows, before the agent takes action elsewhere.
This tool retrieves and queries experiment metadata from Weights & Biases without creating, modifying, deleting, or executing operations. It has no side effects on the system or data. The blast radius of misuse is minimal—an agent could only access information already stored in W&B, not alter it or trigger external actions.
From the tool's definition Tool name 'get_run_details' and description 'Retrieve detailed information about a W&B run' indicate data retrieval with no modification capability.
Attacks that exploit this kind of access
Retrieve detailed information about a W&B run, including:. It is categorised as a Read tool in the Mcp Wandb MCP Server, which means it retrieves data without modifying state.
Register the Mcp Wandb MCP server in PolicyLayer and add a rule for get_run_details: 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 Wandb. Nothing to install.
get_run_details is a Read tool with low risk. Read-only tools are generally safe to allow by default.
Yes. Add a rate_limit block to the get_run_details 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 get_run_details. 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.
get_run_details is provided by the Mcp Wandb MCP server (tsilva/mcp-wandb). 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 →