feat: agent summoning system, todo improvements, security fixes, double-execution bug fix

- Add summon/retrieve tools for spawning child agents in new tabs
  - summon: non-blocking, returns agent_id immediately
  - retrieve: blocking, waits for child to finish, returns result
  - Child tools are intersected with parent permissions (no privilege escalation)
  - Working directory validated to stay within workspace
  - Abort controller stops orphaned agents on tab close
- Rename task_list tool to todo with comprehensive usage guidance in system prompt
- Rename PermissionLog.svelte to ToolPermissions.svelte
- Add 'Summon agents' toggle to tool permissions UI
- Redesign TaskListPanel with DaisyUI checkboxes (indeterminate for in-progress)
- Remove 'blocked' status from task system
- Add tab-created WebSocket event for child agent tab visibility
- Add HMR cleanup for WebSocket connections (close stale connections on hot reload)
- Fix ensureAssistantMessage to not throw on closed tabs
- Fix double tool execution: remove execute from AI SDK tool() in registry.ts
  (agent.ts already executes tools manually via executeToolWithStreaming)
- Fix all pre-existing test failures (missing mocks, stale API signatures)
- Add debug info to copy button (tab ID, injected skills, all tab IDs)
- Add tab ID and tools to conversation copy output

4 files changed, 138 insertions, 15 deletions

diff --git a/packages/core/src/tools/registry.ts b/packages/core/src/tools/registry.ts
index 4699c93..0c7b110 100644
--- a/packages/core/src/tools/registry.ts
+++ b/packages/core/src/tools/registry.ts
@@ -18,15 +18,14 @@ export function createToolRegistry(tools: ToolDefinition[]) {
 			const result: Record<string, ReturnType<typeof tool>> = {};
 			for (const [name, def] of toolMap) {
 				const schema = def.parameters;
+				// Do NOT pass execute here — agent.ts handles tool execution
+				// manually via executeToolWithStreaming. Passing execute would
+				// cause the AI SDK to auto-execute tools AND agent.ts to execute
+				// them again, resulting in double execution.
 				const t = tool({
 					description: def.description,
 					parameters: schema instanceof z.ZodObject ? schema : z.object({}),
-					execute: async (args) => {
-						return def.execute(args as Record<string, unknown>);
-					},
 				});
-				// The AI SDK tool() overloads cause type narrowing issues when
-				// execute is provided. The runtime value is correct.
 				result[name] = t as unknown as ReturnType<typeof tool>;
 			}
 			return result;
diff --git a/packages/core/src/tools/retrieve.ts b/packages/core/src/tools/retrieve.ts
new file mode 100644
index 0000000..93f4c89
--- /dev/null
+++ b/packages/core/src/tools/retrieve.ts
@@ -0,0 +1,41 @@
+import { z } from "zod";
+import type { ToolDefinition } from "../types/index.js";
+
+export interface RetrieveCallbacks {
+	getResult(
+		agentId: string,
+	): Promise<{ status: "done"; result: string } | { status: "error"; error: string }>;
+}
+
+export function createRetrieveTool(callbacks: RetrieveCallbacks): ToolDefinition {
+	return {
+		name: "retrieve",
+		description: [
+			"Wait for a child agent to finish and retrieve its result. This tool BLOCKS until the child completes.",
+			"",
+			"Pass the agent_id returned by the summon tool. Once the child finishes, its final output is returned.",
+			"If the child encountered an error, the error message is returned instead.",
+			"",
+			"Typical usage:",
+			'  1. summon({ task: "...", tools: [...] })  -> get agent_id',
+			"  2. ... do other work or summon more agents ...",
+			'  3. retrieve({ agent_id: "..." })  -> blocks until done, returns result',
+		].join("\n"),
+		parameters: z.object({
+			agent_id: z.string().describe("The agent_id returned by a previous summon call."),
+		}),
+		execute: async (args: Record<string, unknown>): Promise<string> => {
+			const agentId = args.agent_id as string;
+
+			try {
+				const outcome = await callbacks.getResult(agentId);
+				if (outcome.status === "done") {
+					return ["<agent_result>", outcome.result, "</agent_result>"].join("\n");
+				}
+				return `Agent error: ${outcome.error}`;
+			} catch (err) {
+				return `Error retrieving result: ${err instanceof Error ? err.message : String(err)}`;
+			}
+		},
+	};
+}
diff --git a/packages/core/src/tools/summon.ts b/packages/core/src/tools/summon.ts
new file mode 100644
index 0000000..582b871
--- /dev/null
+++ b/packages/core/src/tools/summon.ts
@@ -0,0 +1,88 @@
+import { z } from "zod";
+import type { ToolDefinition } from "../types/index.js";
+
+export interface SummonCallbacks {
+	spawn(options: { task: string; tools: string[]; workingDirectory?: string }): Promise<string>;
+}
+
+export function createSummonTool(
+	defaultWorkingDirectory: string,
+	callbacks: SummonCallbacks,
+): ToolDefinition {
+	return {
+		name: "summon",
+		description: [
+			"Spawn a new child agent to work on a task independently. Returns immediately with an agent_id — does NOT wait for the child to finish.",
+			"",
+			"The child agent runs in its own tab visible to the user. Use the 'retrieve' tool with the returned agent_id to get the result when needed.",
+			"",
+			"Pattern for parallel work:",
+			"  1. Call summon multiple times to start several agents",
+			"  2. Do your own work or wait",
+			"  3. Call retrieve for each agent_id to collect results",
+			"",
+			"The 'tools' parameter controls what the child can do. Available tool names:",
+			"  - read_file: Read file contents",
+			"  - list_files: List files and directories",
+			"  - write_file: Write/edit files",
+			"  - run_shell: Execute shell commands",
+			"  - todo: Track work items",
+			"  - summon: Spawn its own child agents (enables nesting)",
+			"  - retrieve: Collect results from its children (required if summon is given)",
+			"",
+			"If tools is omitted, the child gets read_file, list_files, and todo only (read-only by default).",
+		].join("\n"),
+		parameters: z.object({
+			task: z
+				.string()
+				.describe(
+					"Detailed instructions for the child agent. Be specific about what it should do and what it should return.",
+				),
+			tools: z
+				.array(
+					z.enum([
+						"read_file",
+						"list_files",
+						"write_file",
+						"run_shell",
+						"todo",
+						"summon",
+						"retrieve",
+					]),
+				)
+				.optional()
+				.describe(
+					'Tool names to give the child. Defaults to ["read_file", "list_files", "todo"]. Include "summon" and "retrieve" to allow nesting.',
+				),
+			working_directory: z
+				.string()
+				.optional()
+				.describe(
+					"Absolute path for the child to work in. Defaults to the current working directory.",
+				),
+		}),
+		execute: async (args: Record<string, unknown>): Promise<string> => {
+			const task = args.task as string;
+			const tools = (args.tools as string[] | undefined) ?? ["read_file", "list_files", "todo"];
+			const workingDirectory =
+				(args.working_directory as string | undefined) ?? defaultWorkingDirectory;
+
+			try {
+				const agentId = await callbacks.spawn({
+					task,
+					tools,
+					workingDirectory,
+				});
+				return [
+					`Agent spawned successfully.`,
+					`agent_id: ${agentId}`,
+					``,
+					`The child agent is now working on the task in its own tab.`,
+					`Use the retrieve tool with this agent_id to get the result when ready.`,
+				].join("\n");
+			} catch (err) {
+				return `Error spawning agent: ${err instanceof Error ? err.message : String(err)}`;
+			}
+		},
+	};
+}
diff --git a/packages/core/src/tools/task-list.ts b/packages/core/src/tools/task-list.ts
index 0bacdb4..29f1543 100644
--- a/packages/core/src/tools/task-list.ts
+++ b/packages/core/src/tools/task-list.ts
@@ -60,24 +60,19 @@ export class TaskList {
 
 export function createTaskListTool(taskList: TaskList): ToolDefinition {
 	return {
-		name: "task_list",
+		name: "todo",
 		description:
-			"Manages a task list for tracking work items. The agent can add tasks, update their status, list all tasks, or get details on a specific task.",
+			"Manage a todo list for planning and tracking work. Add items, update their status, list all items, or get details on a specific item.",
 		parameters: z.object({
-			action: z
-				.enum(["add", "update", "list", "get", "remove"])
-				.describe("The action to perform"),
+			action: z.enum(["add", "update", "list", "get", "remove"]).describe("The action to perform"),
 			title: z.string().optional().describe("Task title (required for 'add')"),
 			description: z
 				.string()
 				.optional()
 				.describe("Task description (for 'add', defaults to empty)"),
-			task_id: z
-				.string()
-				.optional()
-				.describe("Task ID (required for 'update', 'get', 'remove')"),
+			task_id: z.string().optional().describe("Task ID (required for 'update', 'get', 'remove')"),
 			status: z
-				.enum(["pending", "in_progress", "done", "blocked"])
+				.enum(["pending", "in_progress", "done"])
 				.optional()
 				.describe("New status (required for 'update')"),
 		}),