The agent determines if a URL fetch has already been performed recently and can reuse the result, avoiding duplicate API calls and saving budget.
The agent prepares a browse request with a specific URL and includes its history of previous tool calls. It sends this to the Agent Preflight MCP server. The server checks if the URL matches a previous successful call within the freshness TTL. If a match is found, the server returns a 'reuse' decision with a result reference. The agent then retrieves the cached result instead of making a new HTTP request. If no match is found, the server returns 'allow', and the agent proceeds to fetch the URL via the companion browse server.
- Category: Operations and infrastructure
- Effect: Read-only
- MCP servers: Agent Preflight MCP (https://koskamo.com/explore/agent-preflight-mcp)
- Source: Agent Preflight MCP documentation (https://api.agent-preflight.com/docs)
- Source quote: "reuse recommends using your prior resultReference; matchedCallId identifies its history entry."
- Playbook page: https://koskamo.com/explore/use-cases/prevent-redundant-browsing-requests
Prompt: Check if I should reuse a previous result for this URL: https://example.com, with history of recent successful fetches.