AI agents invoke cap_compile to trigger actions in Sflight. 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.
Compilation is an Execute-class action because it runs a code generation/transformation process on data. While not destructive or directly modifying the source database, it executes a tool that could theoretically be abused to generate malicious OData metadata or extract sensitive schema information in exportable formats.
From the tool's definition Tool description states it 'Compile[s] CDS model to various output formats' — this is a compilation operation that processes input and generates structured output in multiple formats (json, edm/edmx).
Attacks that exploit this kind of access
Compile CDS model to various output formats: json (CSN), edm/edmx (OData metadata),. It is categorised as a Execute tool in the Sflight MCP Server, which means it can trigger actions or run processes. Use rate limits and argument validation.
Register the Sflight MCP server in PolicyLayer and add a rule for cap_compile: 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 Sflight. Nothing to install.
cap_compile 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 cap_compile 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 cap_compile. 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.
cap_compile is provided by the Sflight MCP server (mindsetconsulting/sflight-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 →