AI agents call find_stack to retrieve information from Unfragile without modifying anything — typically the context-gathering step in research, monitoring, and reporting workflows, before the agent takes action elsewhere.
The tool is querying a graph database of AI capabilities and artifacts to find matching components for a use case. This is fundamentally a Read operation—it retrieves and potentially synthesizes existing data without modifying the database, executing code, or causing side effects. Confidence is slightly reduced due to the truncated description, but the context strongly indicates information retrieval.
From the tool's definition Tool name 'find_stack' and description 'Assemble a complete AI harness stack for a use case' indicates retrieval/assembly of existing artifacts and capabilities.
Attacks that exploit this kind of access
Assemble a complete AI harness stack for a use case. Given a description of what you. It is categorised as a Read tool in the Unfragile MCP Server, which means it retrieves data without modifying state.
Register the Unfragile MCP server in PolicyLayer and add a rule for find_stack: 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 Unfragile. Nothing to install.
find_stack 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 find_stack 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 find_stack. 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.
find_stack is provided by the Unfragile MCP server (savirinc/unfragile-mcp-server). 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 →