search.ai.web
AI-optimized web search — returns synthesized answer + curated results with extracted page content and relevance scores. Built for LLM/agent RAG pipelines. Supports domain filtering and recency (Tavily)
This record as markdown: /tools/io-github-whiteknightonhorse-apibase/search.ai.web.md
What search.ai.web does on Apibase
AI agents call search.ai.web to retrieve information from Apibase without modifying anything. It is typically the context-gathering step in research, monitoring, and reporting workflows, before the agent takes action elsewhere.
| Parameter | Type | Required | Description |
|---|---|---|---|
days | integer | — | Recency filter — only include results from the last N days |
query | string | Yes | Search query — Tavily returns AI-synthesized answer + curated results with extracted content |
max_results | integer | — | Number of results to return (default 5, max 20) |
search_depth | string | — | Search depth: "basic" (faster, 1 credit) or "advanced" (deeper, 2 credits). Default: basic |
include_answer | boolean | — | Include AI-synthesized answer based on search results (default true) |
exclude_domains | array | — | Exclude results from these domains (e.g. ["reddit.com"]) |
include_domains | array | — | Only include results from these domains (e.g. ["wikipedia.org", "arxiv.org"]) |
Parameters from the server's own tool schema.
Why search.ai.web is rated Low
Retrieves and queries public web data with no side effects or state modifications.
From the tool's definition web search, returns synthesized answer, curated results, extracted page content
Attacks that exploit this kind of access
The rule that runs search.ai.web safely
PolicyLayer is an MCP gateway: it sits between your AI agents and Apibase, and checks every tool call against a rule you set before the call runs. Nothing changes on the server itself. For search.ai.web, this is the rule to start with:
search.ai.web is read-only, so it stays allowed. Everything else on the server is denied unless you say otherwise.
The button opens the PolicyLayer dashboard: create your workspace, connect Apibase, apply this rule, and every search.ai.web call is checked against it from then on.
Questions about search.ai.web
AI-optimized web search — returns synthesized answer + curated results with extracted page content and relevance scores. Built for LLM/agent RAG pipelines. Supports domain filtering and recency (Tavily). It is categorised as a Read tool in the Apibase MCP Server, which means it retrieves data without modifying state.
search.ai.web accepts 7 parameters: days, query, max_results, search_depth, include_answer, exclude_domains, include_domains. Required: query. The full parameter table on this page comes from the server's own tool schema.
Register the Apibase MCP server in PolicyLayer and add a rule for search.ai.web: 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 Apibase. Nothing to install.
search.ai.web 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 search.ai.web 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 search.ai.web. 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.
search.ai.web is provided by the Apibase MCP server (apibase-mcp-client). PolicyLayer sits as a proxy in front of this server to enforce policies before tool calls reach the server.
More on Apibase, and thousands of servers like it.
This server
Across the catalogue