agentv 2.0.1 → 2.0.2

package/README.md

@@ -59,6 +59,14 @@ bun run build
 bun test
 ```
 
+5. (Optional) Install example dependencies:
+
+```bash
+bun run examples:install
+```
+
+This step is required if you want to run the examples in the `examples/` directory, as they are self-contained packages with their own dependencies.
+
 You are now ready to start development. The monorepo contains:
 
 - `packages/core/` - Core evaluation engine
@@ -77,9 +85,8 @@ You are now ready to start development. The monorepo contains:
 
 ## Quick Start
 
-You can use the following examples as a starting point.
-- [Simple Example](docs/examples/simple/README.md): A minimal working example to help you get started fast.
-- [Showcase](docs/examples/showcase/README.md): A collection of advanced use cases and real-world agent evaluation scenarios.
+You can use the following examples as a starting point:
+- [Examples](examples/README.md): Feature demonstrations and real-world showcase examples
 
 ### Validating Eval Files
 
@@ -373,7 +380,7 @@ agentv generate rubrics evals/my-eval.yaml --target openai:gpt-4o
   - `borderline`: Score ≥ 0.6 and all required rubrics met
   - `fail`: Score < 0.6 or any required rubric failed
 
-For complete examples and detailed patterns, see [examples/features/evals/rubric/](examples/features/evals/rubric/).
+For complete examples and detailed patterns, see [examples/features/rubric/](examples/features/rubric/).
 
 ## Advanced Configuration
 

package/dist/{chunk-6SHT2QS6.js → chunk-5AJ7DFUO.js}

@@ -375,7 +375,7 @@ var compareCommand = command({
 import { readFileSync as readFileSync2, writeFileSync } from "node:fs";
 import path15 from "node:path";
 
-// ../../packages/core/dist/chunk-IBTKEEOT.js
+// ../../packages/core/dist/chunk-KDEP4I7G.js
 import { constants } from "node:fs";
 import { access, readFile } from "node:fs/promises";
 import path from "node:path";
@@ -4422,7 +4422,7 @@ var coerce = {
 };
 var NEVER = INVALID;
 
-// ../../packages/core/dist/chunk-IBTKEEOT.js
+// ../../packages/core/dist/chunk-KDEP4I7G.js
 async function fileExists(filePath) {
   try {
     await access(filePath, constants.F_OK);
@@ -4802,6 +4802,15 @@ function resolveTargetDefinition(definition, env = process.env, evalFilePath) {
         providerBatching,
         config: resolvePiCodingAgentConfig(parsed, env)
       };
+    case "pi-agent-sdk":
+      return {
+        kind: "pi-agent-sdk",
+        name: parsed.name,
+        judgeTarget: parsed.judge_target,
+        workers: parsed.workers,
+        providerBatching,
+        config: resolvePiAgentSdkConfig(parsed, env)
+      };
     case "claude-code":
       return {
         kind: "claude-code",
@@ -5023,6 +5032,39 @@ function resolvePiCodingAgentConfig(target, env) {
     systemPrompt
   };
 }
+function resolvePiAgentSdkConfig(target, env) {
+  const providerSource = target.pi_provider ?? target.piProvider ?? target.llm_provider;
+  const modelSource = target.model ?? target.pi_model ?? target.piModel;
+  const apiKeySource = target.api_key ?? target.apiKey;
+  const timeoutSource = target.timeout_seconds ?? target.timeoutSeconds;
+  const systemPromptSource = target.system_prompt ?? target.systemPrompt;
+  const provider = resolveOptionalString(
+    providerSource,
+    env,
+    `${target.name} pi-agent-sdk provider`,
+    {
+      allowLiteral: true,
+      optionalEnv: true
+    }
+  );
+  const model = resolveOptionalString(modelSource, env, `${target.name} pi-agent-sdk model`, {
+    allowLiteral: true,
+    optionalEnv: true
+  });
+  const apiKey = resolveOptionalString(apiKeySource, env, `${target.name} pi-agent-sdk api key`, {
+    allowLiteral: false,
+    optionalEnv: true
+  });
+  const timeoutMs = resolveTimeoutMs(timeoutSource, `${target.name} pi-agent-sdk timeout`);
+  const systemPrompt = typeof systemPromptSource === "string" && systemPromptSource.trim().length > 0 ? systemPromptSource.trim() : void 0;
+  return {
+    provider,
+    model,
+    apiKey,
+    timeoutMs,
+    systemPrompt
+  };
+}
 function resolveClaudeCodeConfig(target, env) {
   const executableSource = target.executable ?? target.command ?? target.binary;
   const modelSource = target.model;
@@ -5331,6 +5373,7 @@ var KNOWN_PROVIDERS = [
   "gemini",
   "codex",
   "pi-coding-agent",
+  "pi-agent-sdk",
   "claude-code",
   "cli",
   "mock",
@@ -38948,6 +38991,165 @@ var MockProvider = class {
     return this.delayMs;
   }
 };
+var piAgentModule = null;
+var piAiModule = null;
+async function loadPiModules() {
+  if (!piAgentModule || !piAiModule) {
+    try {
+      [piAgentModule, piAiModule] = await Promise.all([
+        import("@mariozechner/pi-agent"),
+        import("@mariozechner/pi-ai")
+      ]);
+    } catch (error40) {
+      throw new Error(
+        `Failed to load pi-agent-sdk dependencies. Please install them:
+  npm install @mariozechner/pi-agent @mariozechner/pi-ai
+
+Original error: ${error40 instanceof Error ? error40.message : String(error40)}`
+      );
+    }
+  }
+  return {
+    Agent: piAgentModule.Agent,
+    ProviderTransport: piAgentModule.ProviderTransport,
+    getModel: piAiModule.getModel,
+    getEnvApiKey: piAiModule.getEnvApiKey
+  };
+}
+var PiAgentSdkProvider = class {
+  id;
+  kind = "pi-agent-sdk";
+  targetName;
+  supportsBatch = false;
+  config;
+  constructor(targetName, config2) {
+    this.id = `pi-agent-sdk:${targetName}`;
+    this.targetName = targetName;
+    this.config = config2;
+  }
+  async invoke(request) {
+    if (request.signal?.aborted) {
+      throw new Error("Pi agent SDK request was aborted before execution");
+    }
+    const { Agent, ProviderTransport, getModel, getEnvApiKey } = await loadPiModules();
+    const startTime = Date.now();
+    const providerName = this.config.provider ?? "anthropic";
+    const modelId = this.config.model ?? "claude-sonnet-4-20250514";
+    const model = getModel(providerName, modelId);
+    const systemPrompt = this.config.systemPrompt ?? "Answer directly and concisely.";
+    const transport = new ProviderTransport({
+      getApiKey: async (provider) => {
+        return this.config.apiKey ?? getEnvApiKey(provider) ?? void 0;
+      }
+    });
+    const agent = new Agent({
+      initialState: {
+        systemPrompt,
+        model,
+        tools: [],
+        // No tools for simple Q&A
+        messages: []
+      },
+      transport
+    });
+    const outputMessages = [];
+    let finalAssistantContent = "";
+    const unsubscribe = agent.subscribe((event) => {
+      if (event.type === "message_end") {
+        const msg = event.message;
+        if (msg.role === "assistant") {
+          const content = extractTextContent22(msg.content);
+          if (content) {
+            finalAssistantContent = content;
+          }
+        }
+      }
+    });
+    try {
+      const timeoutMs = this.config.timeoutMs ?? 12e4;
+      const timeoutPromise = new Promise((_, reject) => {
+        setTimeout(
+          () => reject(new Error(`Pi agent SDK timed out after ${timeoutMs}ms`)),
+          timeoutMs
+        );
+      });
+      await Promise.race([agent.prompt(request.question), timeoutPromise]);
+      await agent.waitForIdle();
+      const agentMessages = agent.state.messages;
+      for (const msg of agentMessages) {
+        outputMessages.push(convertAgentMessage(msg));
+      }
+      const durationMs = Date.now() - startTime;
+      return {
+        raw: {
+          messages: agentMessages,
+          systemPrompt,
+          model: this.config.model,
+          provider: this.config.provider
+        },
+        outputMessages,
+        durationMs
+      };
+    } finally {
+      unsubscribe();
+    }
+  }
+};
+function extractTextContent22(content) {
+  if (typeof content === "string") {
+    return content;
+  }
+  if (!Array.isArray(content)) {
+    return void 0;
+  }
+  const textParts = [];
+  for (const part of content) {
+    if (!part || typeof part !== "object") {
+      continue;
+    }
+    const p = part;
+    if (p.type === "text" && typeof p.text === "string") {
+      textParts.push(p.text);
+    }
+  }
+  return textParts.length > 0 ? textParts.join("\n") : void 0;
+}
+function convertAgentMessage(message) {
+  if (!message || typeof message !== "object") {
+    return { role: "unknown", content: String(message) };
+  }
+  const msg = message;
+  const role = typeof msg.role === "string" ? msg.role : "unknown";
+  const content = extractTextContent22(msg.content);
+  const toolCalls = extractToolCalls2(msg.content);
+  const timestamp = typeof msg.timestamp === "number" ? new Date(msg.timestamp).toISOString() : typeof msg.timestamp === "string" ? msg.timestamp : void 0;
+  return {
+    role,
+    content,
+    toolCalls: toolCalls.length > 0 ? toolCalls : void 0,
+    timestamp
+  };
+}
+function extractToolCalls2(content) {
+  if (!Array.isArray(content)) {
+    return [];
+  }
+  const toolCalls = [];
+  for (const part of content) {
+    if (!part || typeof part !== "object") {
+      continue;
+    }
+    const p = part;
+    if (p.type === "tool_use" && typeof p.name === "string") {
+      toolCalls.push({
+        tool: p.name,
+        input: p.input,
+        id: typeof p.id === "string" ? p.id : void 0
+      });
+    }
+  }
+  return toolCalls;
+}
 var GLOBAL_LOGS_KEY3 = Symbol.for("agentv.piLogs");
 var GLOBAL_SUBSCRIBERS_KEY3 = Symbol.for("agentv.piLogSubscribers");
 function getPiLogStore() {
@@ -39445,8 +39647,8 @@ function convertPiMessage(message) {
   if (typeof role !== "string") {
     return void 0;
   }
-  const content = extractTextContent22(msg.content);
-  const toolCalls = extractToolCalls2(msg.content);
+  const content = extractTextContent3(msg.content);
+  const toolCalls = extractToolCalls3(msg.content);
   const timestamp = typeof msg.timestamp === "number" ? new Date(msg.timestamp).toISOString() : typeof msg.timestamp === "string" ? msg.timestamp : void 0;
   const metadata = {};
   if (msg.api) metadata.api = msg.api;
@@ -39462,7 +39664,7 @@ function convertPiMessage(message) {
     metadata: Object.keys(metadata).length > 0 ? metadata : void 0
   };
 }
-function extractTextContent22(content) {
+function extractTextContent3(content) {
   if (typeof content === "string") {
     return content;
   }
@@ -39481,7 +39683,7 @@ function extractTextContent22(content) {
   }
   return textParts.length > 0 ? textParts.join("\n") : void 0;
 }
-function extractToolCalls2(content) {
+function extractToolCalls3(content) {
   if (!Array.isArray(content)) {
     return [];
   }
@@ -39955,6 +40157,8 @@ function createProvider(target) {
       return new CodexProvider(target.name, target.config);
     case "pi-coding-agent":
       return new PiCodingAgentProvider(target.name, target.config);
+    case "pi-agent-sdk":
+      return new PiAgentSdkProvider(target.name, target.config);
     case "claude-code":
       return new ClaudeCodeProvider(target.name, target.config);
     case "mock":
@@ -45694,4 +45898,4 @@ export {
   app,
   runCli
 };
-//# sourceMappingURL=chunk-6SHT2QS6.js.map
+//# sourceMappingURL=chunk-5AJ7DFUO.js.map