Remove Tavily websearch built-in tool

README.md

@@ -86,7 +86,7 @@ In another terminal:
 
 ```sh
 pnpm dlx @restatedev/restate deployments register http://localhost:9080
-curl localhost:8080/Hello/ask --json '"Say hello in one sentence."'
+curl localhost:8080/Hello/ask --json '{"message":"Say hello in one sentence."}'
 ```
 
 TypeScript callers can use Restate's typed client instead, as the
@@ -98,7 +98,7 @@ import {hello} from "@restatedev/agentlib-examples/hello";
 
 const answer = await connect({url: "http://localhost:8080"})
   .client(hello)
-  .ask("Say hello in one sentence.");
+  .ask({message: "Say hello in one sentence."});
 console.log(answer);
 ```
 
@@ -157,7 +157,7 @@ repeat a completed model call.
 | MCP                          | `mcpTools` (`/mcp`), passed as `run(input, {tools})`                       | [17](packages/examples/src/lessons/17_mcp_tools.ts)         |
 | AG-UI clients                | `onProgress`, `onApproval` and client-run tools as AG-UI events            | [18](packages/examples/src/lessons/18_ag_ui.ts)             |
 | Guardrails                   | `beforeRun`, `beforeStep`, `beforeTool`, `afterTool`, `afterRun`, `before` | [13](packages/examples/src/lessons/13_tool_guardrail.ts)    |
-| Web search                   | `webSearch` (`/web-search`)                                                | [14](packages/examples/src/lessons/14_web_search.ts)        |
+| Web search                   | Example-local `asyncTool` calling Tavily                                   | [14](packages/examples/src/lessons/14_web_search.ts)        |
 | Skills                       | `agent({skills})` with any `SkillSource`, `skills([...])`                  | [19](packages/examples/src/lessons/19_skills.ts)            |
 | Large tool results           | `agent({toolResults})`: previews, read on with `read_tool_result`          | —                                                           |
 | Long runs                    | `agent({compaction, contextWindow})`, `ContextOverflowError`               | —                                                           |

packages/core/README.md

@@ -37,15 +37,14 @@ go from a first run to multi-agent workflows.
 
 ## Subpaths
 
-| Import                            | Purpose                                       |
-| --------------------------------- | --------------------------------------------- |
-| `@restatedev/agentlib`            | Agents, tools, conversations, guardrails      |
-| `@restatedev/agentlib/openai`     | OpenAI and Chat Completions-compatible models |
-| `@restatedev/agentlib/provider`   | Models through the Vercel AI SDK providers    |
-| `@restatedev/agentlib/mcp`        | Tools from MCP servers                        |
-| `@restatedev/agentlib/program`    | Programmatic tool calling in a sandbox        |
-| `@restatedev/agentlib/web-search` | Web search tools                              |
-| `@restatedev/agentlib/discovery`  | Expose agents and tools for discovery         |
+| Import                           | Purpose                                       |
+| -------------------------------- | --------------------------------------------- |
+| `@restatedev/agentlib`           | Agents, tools, conversations, guardrails      |
+| `@restatedev/agentlib/openai`    | OpenAI and Chat Completions-compatible models |
+| `@restatedev/agentlib/provider`  | Models through the Vercel AI SDK providers    |
+| `@restatedev/agentlib/mcp`       | Tools from MCP servers                        |
+| `@restatedev/agentlib/program`   | Programmatic tool calling in a sandbox        |
+| `@restatedev/agentlib/discovery` | Expose agents and tools for discovery         |
 
 ## License
 

packages/core/package.json

@@ -45,10 +45,6 @@
       "types": "./dist/provider.d.ts",
       "import": "./dist/provider.js"
     },
-    "./web-search": {
-      "types": "./dist/web-search.d.ts",
-      "import": "./dist/web-search.js"
-    },
     "./discovery": {
       "types": "./dist/discovery.d.ts",
       "import": "./dist/discovery.js"

packages/examples/README.md

@@ -10,8 +10,8 @@ the services in tutorial order, [app.ts](src/app.ts) serves them, and
 Each prompt handler declares a Zod input schema with a scenario-specific
 default for the request editor. The research example also supplies demo sources. Rediscover
 the endpoint after changing a schema so Restate receives the updated defaults.
-String bodies use `.meta({default: ...})` to prefill the editor while keeping
-the request body required; object fields use `.default(...)`.
+Prompt inputs use JSON objects with `.default(...)` on their fields. This keeps
+string values correctly quoted in the Restate UI request editor, ready to send.
 
 | Lesson | New concept                           | Source                                                             |
 | ------ | ------------------------------------- | ------------------------------------------------------------------ |
@@ -46,7 +46,7 @@ If you have a specific goal, start here:
 | Steer or interrupt a running agent   | [09](src/lessons/09_support_session.ts) |
 | Route a request by its meaning       | [12](src/lessons/12_routing.ts)         |
 | Enforce a budget before an action    | [13](src/lessons/13_tool_guardrail.ts)  |
-| Answer with current public sources   | [14](src/lessons/14_web_search.ts)      |
+| Answer with public sources           | [14](src/lessons/14_web_search.ts)      |
 | Let agents review each other\'s work | [15](src/lessons/15_peer_review_rpc.ts) |
 | Show progress while an agent works   | [16](src/lessons/16_progress_feed.ts)   |
 | Use tools from an MCP server         | [17](src/lessons/17_mcp_tools.ts)       |
@@ -86,7 +86,7 @@ to the agent and returns its answer as a JSON string. Each call starts with no
 conversation history.
 
 ```sh
-curl --globoff "$RESTATE_CALL_URL/Hello/ask" --json '"Say hello in one sentence."'
+curl --globoff "$RESTATE_CALL_URL/Hello/ask" --json '{"message":"Say hello in one sentence."}'
 ```
 
 The response is just the answer, such as `"Hello!"`. A run that stopped or was
@@ -105,7 +105,7 @@ result in its answer.
 
 ```sh
 curl --globoff "$RESTATE_CALL_URL/TicketLookup/ask" \
-  --json '"What is the status of ticket T-42?"'
+  --json '{"message":"What is the status of ticket T-42?"}'
 ```
 
 The lesson uses `asyncTool` with an ordinary `async execute(...)` callback.
@@ -127,8 +127,8 @@ messages. Pass them to the agent, then save `result.messages` afterward. Return
 only the reply; the history stays in object state.
 
 ```sh
-curl --globoff "$RESTATE_CALL_URL/ChatSession/bob/ask" --json '"My name is Bob."'
-curl --globoff "$RESTATE_CALL_URL/ChatSession/bob/ask" --json '"What is my name?"'
+curl --globoff "$RESTATE_CALL_URL/ChatSession/bob/ask" --json '{"message":"My name is Bob."}'
+curl --globoff "$RESTATE_CALL_URL/ChatSession/bob/ask" --json '{"message":"What is my name?"}'
 ```
 
 Both requests use `bob`, so the second sees the first exchange. A different key gets
@@ -142,7 +142,7 @@ answer and gives the caller the typed object directly.
 
 ```sh
 curl --globoff "$RESTATE_CALL_URL/TicketTriage/classify" \
-  --json '"I was charged twice for my subscription."'
+  --json '{"message":"I was charged twice for my subscription."}'
 ```
 
 A successful response is an object such as
@@ -180,7 +180,7 @@ Start a request in one terminal:
 
 ```sh
 curl --globoff "$RESTATE_CALL_URL/TicketApproval/bob/ask" \
-  --json '"Close ticket T-42 after I approve it."'
+  --json '{"message":"Close ticket T-42 after I approve it."}'
 ```
 
 This request waits. Before executing the tool, the SDK calls `onApproval(request)`
@@ -293,7 +293,7 @@ to one handler instead. Virtual objects and workflows also take a `key`.
 
 ```sh
 curl --globoff "$RESTATE_CALL_URL/RefundAgent/ask" \
-  --json '"Refund order O-99 for $12.50 and tell me the refund policy."'
+  --json '{"message":"Refund order O-99 for $12.50 and tell me the refund policy."}'
 ```
 
 Each call is a durable child invocation. Interrupting the run cancels an
@@ -331,7 +331,7 @@ Authenticate with `modal token new`, or set `MODAL_TOKEN_ID` and
 
 ```sh
 curl --globoff "$RESTATE_CALL_URL/CodingAgent/ask" \
-  --json '"Write primes.js that prints the first 10 primes, then run it."'
+  --json '{"task":"Write primes.js that prints the first 10 primes, then run it."}'
 ```
 
 ## 12 — Route a request to a specialist
@@ -351,7 +351,7 @@ function* (question) {
 
 ```sh
 curl --globoff "$RESTATE_CALL_URL/SupportRouter/answer" \
-  --json '"I cannot log into my account. How can I regain access?"'
+  --json '{"question":"I cannot log into my account. How can I regain access?"}'
 ```
 
 The specialist starts a fresh run with only the question. Use `handoff` when
@@ -381,30 +381,24 @@ Raise `maxTotalCents` to `15000` to see the allowed path. The reservation is a
 demo result; connect a real booking API only after obtaining the quote from your
 own trusted service.
 
-## 14 — Answer with current public sources
+## 14 — Search the web with a custom tool
 
-`webSearch` sends a public query to Tavily and journals the bounded results, so
-a replay does not repeat the HTTP request. Without an API key it uses Tavily's
-free, rate-limited keyless access; pass `webSearch({apiKey})` for higher limits. The agent is instructed to cite returned URLs and treat
-snippets as evidence, not instructions.
+[14_web_search.ts](src/lessons/14_web_search.ts) defines `searchWeb` locally with
+`asyncTool`. Its `execute` function calls the [Tavily Search API](https://docs.tavily.com/documentation/api-reference/endpoint/search)
+using `fetch`, validates the response, and returns up to five bounded snippets.
+`asyncTool` journals the result, so replay does not repeat a completed search.
+This is an example of integrating an API, not a built-in library tool.
 
-```ts
-const researcher = agent({
-  model,
-  maxSteps: 4,
-  instructions: "Search the web before answering. Cite URLs returned by searchWeb, ...",
-  tools: {searchWeb: webSearch()},
-});
-```
+Set `TAVILY_API_KEY` in the example server's environment before starting it.
+The key is read inside `execute` and is never included in the model's tool arguments.
+Keep private data out of queries; the agent treats snippets as untrusted evidence
+and cites the returned URLs.
 
 ```sh
 curl --globoff "$RESTATE_CALL_URL/WebResearch/answer" \
-  --json '"Find the current Restate TypeScript SDK docs and cite their URLs."'
+  --json '{"question":"Find the Restate TypeScript SDK docs and cite their URLs."}'
 ```
 
-Keep private data out of search queries. A key, when you pass one, stays in
-application code and is not part of the model's tool schema.
-
 ## 15 — Let agents exchange feedback over RPC
 
 `PeerReview` asks `PeerAuthor` for a draft, sends that draft to two
@@ -432,7 +426,7 @@ sequenceDiagram
 
 ```sh
 curl --globoff "$RESTATE_CALL_URL/PeerReview/answer" \
-  --json '"Write a two-sentence release note for a feature that retries transient failures and resumes completed steps from a journal."'
+  --json '{"brief":"Write a two-sentence release note for a feature that retries transient failures and resumes completed steps from a journal."}'
 ```
 
 The caller receives only the revised answer. Restate journals each RPC, so a
@@ -456,7 +450,7 @@ each tool into a journaled step. The lesson gives an agent the tools of the
 Restate docs server at `https://docs.restate.dev/mcp`.
 
 ```sh
-curl "$RESTATE_CALL_URL/DocsGuide/ask" --json '"What is a virtual object?"'
+curl "$RESTATE_CALL_URL/DocsGuide/ask" --json '{"question":"What is a virtual object?"}'
 ```
 
 ## 18 — Serve an AG-UI client
@@ -514,7 +508,7 @@ Skills defined in code need no folder: use `skills([{name, description,
 instructions, files}])` from `@restatedev/agentlib`.
 
 ```sh
-curl "$RESTATE_CALL_URL/SkilledWriter/write" --json '"Release notes for 2.4.0: added CSV export, fixed the login redirect loop, bumped lodash."'
+curl "$RESTATE_CALL_URL/SkilledWriter/write" --json '{"request":"Release notes for 2.4.0: added CSV export, fixed the login redirect loop, bumped lodash."}'
 ```
 
 A skill's scripts are out of scope: the SDK offers no shell to run them. An
@@ -598,9 +592,8 @@ commit `60d1eda2e5f3aae96db2efeba498e52367a01648`:
 | Racing agents   | [racing-agents.ts](https://github.com/restatedev/ai-examples/blob/60d1eda2e5f3aae96db2efeba498e52367a01648/typescript-restate-only/tour-of-agents/src/racing-agents.ts)                               | Embedded operations, first completed success, cancellation with joined cleanup                         |
 
 The later goal-based lessons draw on the same patterns described by
-[Anthropic's workflow guide](https://www.anthropic.com/engineering/building-effective-agents),
-[OpenAI's tool guardrails](https://openai.github.io/openai-agents-python/guardrails/),
-and the [AI SDK web-search recipe](https://ai-sdk.dev/cookbook/node/web-search-agent).
+[Anthropic's workflow guide](https://www.anthropic.com/engineering/building-effective-agents)
+and [OpenAI's tool guardrails](https://openai.github.io/openai-agents-python/guardrails/).
 They use this library's Restate handlers and tools rather than copying another
 framework's control flow.
 

packages/examples/src/lessons/01_hello.ts

@@ -10,8 +10,11 @@ export const hello = restate.service({
   name: "Hello",
   handlers: {
     ask: restate.schemas(
-      {input: z.string().min(1).meta({default: "Say hello in one sentence."}), output: z.string()},
-      function* (message) {
+      {
+        input: z.object({message: z.string().default("Say hello in one sentence.")}),
+        output: z.string(),
+      },
+      function* ({message}) {
         const result = yield* assistant.run(message);
         if (result.status !== "completed") return result.output ?? result.reason;
         return result.output;

packages/examples/src/lessons/02_tools.ts

@@ -31,10 +31,10 @@ export const ticketLookup = restate.service({
   handlers: {
     ask: restate.schemas(
       {
-        input: z.string().min(1).meta({default: "What is the status of ticket T-42?"}),
+        input: z.object({message: z.string().default("What is the status of ticket T-42?")}),
         output: z.string(),
       },
-      function* (message) {
+      function* ({message}) {
         const result = yield* support.run(message);
         if (result.status !== "completed") return result.output ?? result.reason;
         return result.output;

packages/examples/src/lessons/03_conversation.ts

@@ -14,13 +14,15 @@ export const chatSession = restate.object({
   handlers: {
     ask: restate.schemas(
       {
-        input: z
-          .string()
-          .min(1)
-          .meta({default: "My name is Bob. Remember it for our next conversation."}),
+        input: z.object({
+          message: z
+            .string()
+
+            .default("My name is Bob. Remember it for our next conversation."),
+        }),
         output: z.string(),
       },
-      function* (message) {
+      function* ({message}) {
         const messages = (yield* restate.state().get<Message[]>("messages")) ?? [];
         const result = yield* assistant.run(message, {messages});
         restate.state().set("messages", result.messages);

packages/examples/src/lessons/04_structured_output.ts

@@ -21,13 +21,15 @@ export const ticketTriage = restate.service({
   handlers: {
     classify: restate.schemas(
       {
-        input: z
-          .string()
-          .min(1)
-          .meta({default: "I was charged twice for my subscription. Please help me get a refund."}),
+        input: z.object({
+          message: z
+            .string()
+
+            .default("I was charged twice for my subscription. Please help me get a refund."),
+        }),
         output: classificationSchema,
       },
-      function* (message) {
+      function* ({message}) {
         const result = yield* triage.run(message);
         if (result.status !== "completed")
           throw new TerminalError(`Triage ${result.status}: ${result.reason}`);

packages/examples/src/lessons/05_evaluator_optimizer.ts

@@ -7,15 +7,15 @@ import {z} from "zod";
 import {exampleModel as model} from "../shared.js";
 
 const MAX_ITERATIONS = 3;
-const evaluationSchema = z.object({approved: z.boolean(), feedback: z.string().min(1).max(4000)});
+const evaluationSchema = z.object({approved: z.boolean(), feedback: z.string().max(4000)});
 
 const generator = agent({
   name: "code-writer",
   model,
   maxSteps: 3,
   instructions:
     "Code writer. Write a small, self-contained TypeScript solution to the task. When a previous draft and feedback are supplied, correct that draft. Return code as data; nothing is executed.",
-  output: z.object({code: z.string().min(1).max(24000)}),
+  output: z.object({code: z.string().max(24000)}),
 });
 const evaluator = agent({
   name: "code-reviewer",
@@ -33,7 +33,7 @@ export const codeReview = restate.service({
         input: z.object({
           task: z
             .string()
-            .min(1)
+
             .max(4000)
             .default(
               "Write a TypeScript function that returns the first item of an array, or undefined when it is empty.",