orchajs 0.6.2 → 0.7.1

package/dist/runtime/agent.js

@@ -1,5 +1,6 @@
 import { randomUUID } from "node:crypto";
 import { OrchaError } from "../errors.js";
+import { normalizeLocalFile, normalizeUrlContent, } from "../file-content.js";
 import { messagesFromEvents, normalizePersistedToolResults, SessionEventWriter, } from "./session-events.js";
 import { validateStructuredValue } from "./structured-output.js";
 import { executeLocalAction } from "./actions/execute.js";
@@ -25,6 +26,7 @@ export class RuntimeAgent {
     #delegatedOnly;
     #agentKey;
     #sessionIdPrefix;
+    #projectRoot;
     #evaluationTasks = new Map();
     #sessionIdleWaiters = new Map();
     constructor(options) {
@@ -38,6 +40,7 @@ export class RuntimeAgent {
         this.#delegatedOnly = options.delegatedOnly ?? false;
         this.#agentKey = options.manifest.key ?? options.manifest.name;
         this.#sessionIdPrefix = options.sessionIdPrefix ?? "ses_";
+        this.#projectRoot = options.projectRoot ?? process.cwd();
     }
     get agent() {
         return this.#agentKey;
@@ -379,7 +382,13 @@ export class RuntimeAgent {
         return [...events, ...recoveryEvents];
     }
     async #start(sessionId, input, createSession = true, publishOutput, publishSnapshot, lineage) {
-        const content = normalizeContent(input.content);
+        let content;
+        try {
+            content = normalizeContent(input.content, this.#projectRoot);
+        }
+        catch (error) {
+            return failure(sessionId, "invalid_input", error instanceof Error ? error.message : String(error));
+        }
         if (!content) {
             return failure(sessionId, "invalid_input", "Agent content must contain at least one valid item.");
         }
@@ -1040,8 +1049,8 @@ export class RuntimeAgent {
                             description: "The registered subagent to run.",
                         },
                         input: {
-                            type: "string",
-                            description: "The complete task and relevant context for the subagent.",
+                            ...agentContentInputSchema(),
+                            description: "The complete text or multimodal content for the subagent. Local file paths resolve from the configured Orcha project root.",
                         },
                     },
                     required: ["agent", "input"],
@@ -1123,15 +1132,16 @@ export class RuntimeAgent {
     async #executeSubagentToolCall(parentSessionId, writer, call, events, publishSnapshot) {
         if (call.name === RUN_AGENT_TOOL_NAME) {
             const agentName = requiredToolString(call.arguments.agent, "agent");
-            const input = requiredToolString(call.arguments.input, "input");
             const child = this.#subagents[agentName];
             if (!child) {
                 throw new Error(`Unknown subagent "${agentName}".`);
             }
-            const existing = previousSubagentCall(await this.#readSession(parentSessionId), call.callId);
+            const parentEvents = await this.#readSession(parentSessionId);
+            const existing = previousSubagentCall(parentEvents, call.callId);
             if (existing) {
                 return subagentToolOutput(existing);
             }
+            const input = normalizeDelegatedInput(call.arguments.input, this.#projectRoot);
             const lineage = {
                 origin: "delegated",
                 parentAgent: this.#manifest.key,
@@ -1147,7 +1157,7 @@ export class RuntimeAgent {
             });
             await this.#store.append(parentSessionId, [initiated]);
             events.push(initiated);
-            const childExecution = child.runAsChild(childSessionId, input, lineage);
+            const childExecution = child.runAsChild(childSessionId, { content: input }, lineage);
             publishSnapshot?.({
                 sessionId: parentSessionId,
                 status: "waiting_for_subagent",
@@ -1621,24 +1631,49 @@ function isEvaluationMetricResult(value) {
         Array.isArray(value.evidence) &&
         value.evidence.every((item) => typeof item === "string"));
 }
-function normalizeContent(content) {
+function normalizeContent(content, projectRoot) {
     if (typeof content === "string") {
         return content.trim()
             ? [{ type: "text", text: content }]
             : undefined;
     }
-    if (!Array.isArray(content) || content.length === 0) {
+    const items = Array.isArray(content)
+        ? content
+        : [content];
+    if (items.length === 0) {
         return undefined;
     }
     const normalized = [];
-    for (const item of content) {
-        if (item?.type === "text" &&
+    for (const item of items) {
+        if (item &&
+            "type" in item &&
+            item.type === "text" &&
             typeof item.text === "string" &&
             item.text.trim()) {
             normalized.push({ type: "text", text: item.text });
             continue;
         }
         if (item &&
+            "filePath" in item &&
+            typeof item.filePath === "string" &&
+            item.filePath.trim()) {
+            normalized.push(normalizeLocalFile(item.filePath, "mimeType" in item && typeof item.mimeType === "string"
+                ? item.mimeType
+                : undefined, projectRoot));
+            continue;
+        }
+        if (item &&
+            "url" in item &&
+            typeof item.url === "string" &&
+            item.url.trim()) {
+            normalized.push(normalizeUrlContent(item.url, "mimeType" in item && typeof item.mimeType === "string"
+                ? item.mimeType
+                : undefined));
+            continue;
+        }
+        if (item &&
+            "type" in item &&
+            typeof item.type === "string" &&
             ["image", "video", "audio", "url"].includes(item.type) &&
             "mimeType" in item &&
             "fileUri" in item &&
@@ -1653,6 +1688,82 @@ function normalizeContent(content) {
     }
     return normalized;
 }
+function normalizeDelegatedInput(input, projectRoot) {
+    if (typeof input === "string") {
+        if (!input.trim()) {
+            throw new Error('run_agent requires a non-empty "input".');
+        }
+        return input;
+    }
+    if (!isRecord(input) && !Array.isArray(input)) {
+        throw new Error("run_agent input must be text, one content item, or a content array.");
+    }
+    const content = normalizeContent(input, projectRoot);
+    if (!content) {
+        throw new Error("run_agent input must contain at least one valid content item.");
+    }
+    return content;
+}
+function agentContentInputSchema() {
+    const textItem = {
+        type: "object",
+        properties: {
+            type: { type: "string", enum: ["text"] },
+            text: { type: "string", minLength: 1 },
+        },
+        required: ["type", "text"],
+        additionalProperties: false,
+    };
+    const fileItems = [
+        {
+            type: "object",
+            properties: {
+                type: { type: "string", enum: ["file"] },
+                filePath: { type: "string", minLength: 1 },
+                mimeType: { type: "string", minLength: 1 },
+            },
+            required: ["filePath"],
+            additionalProperties: false,
+        },
+        {
+            type: "object",
+            properties: {
+                type: { type: "string", enum: ["url"] },
+                url: { type: "string", minLength: 1 },
+                mimeType: { type: "string", minLength: 1 },
+            },
+            required: ["url"],
+            additionalProperties: false,
+        },
+        {
+            type: "object",
+            properties: {
+                type: {
+                    type: "string",
+                    enum: ["image", "video", "audio", "url"],
+                },
+                fileUri: { type: "string", minLength: 1 },
+                mimeType: { type: "string", minLength: 1 },
+            },
+            required: ["type", "fileUri", "mimeType"],
+            additionalProperties: false,
+        },
+    ];
+    const item = {
+        anyOf: [textItem, ...fileItems],
+    };
+    return {
+        anyOf: [
+            { type: "string", minLength: 1 },
+            item,
+            {
+                type: "array",
+                items: item,
+                minItems: 1,
+            },
+        ],
+    };
+}
 function normalizeClientCapabilities(capabilities) {
     const names = [];
     for (const capability of capabilities) {
@@ -1750,7 +1861,8 @@ function buildSystemPrompt(manifest, events, variables) {
     if (subagents.length > 0) {
         sections.push([
             "## Available subagents",
-            "Use `run_agent` to start a specialist. Give it the complete task and only the context it needs.",
+            "Use `run_agent` to start a specialist. Give it the complete task and only the context it needs. Its input accepts the same text and multimodal content format as a normal agent run.",
+            "Local file paths in subagent input resolve from the configured Orcha project root.",
             "You may request multiple independent subagents in one response; Orcha runs them concurrently and returns each result by call ID.",
             "The call waits synchronously while the child is actively running, and the child's text response returns as a normal tool result.",
             "If a child pauses for a client action or because its session was paused, decide whether to resume it with `resume_agent`, inspect it with `inspect_agent`, invoke one of your own client actions for missing information, or continue without it.",
@@ -2281,6 +2393,10 @@ function isMessageContent(value) {
     if (value.type === "text") {
         return typeof value.text === "string";
     }
+    if (value.type === "file") {
+        return (typeof value.mimeType === "string" &&
+            typeof value.filePath === "string");
+    }
     return (["image", "video", "audio", "url"].includes(value.type) &&
         typeof value.mimeType === "string" &&
         typeof value.fileUri === "string");