@oh-my-pi/pi-coding-agent 15.13.1 → 15.13.3

package/package.json

@@ -1,7 +1,7 @@
 {
 	"type": "module",
 	"name": "@oh-my-pi/pi-coding-agent",
-	"version": "15.13.1",
+	"version": "15.13.3",
 	"description": "Coding agent CLI with read, bash, edit, write tools and session management",
 	"homepage": "https://omp.sh",
 	"author": "Can Boluk",
@@ -47,17 +47,17 @@
 		"@agentclientprotocol/sdk": "0.25.0",
 		"@babel/parser": "^7.29.7",
 		"@mozilla/readability": "^0.6.0",
-		"@oh-my-pi/hashline": "15.13.1",
-		"@oh-my-pi/omp-stats": "15.13.1",
-		"@oh-my-pi/pi-agent-core": "15.13.1",
-		"@oh-my-pi/pi-ai": "15.13.1",
-		"@oh-my-pi/pi-catalog": "15.13.1",
-		"@oh-my-pi/pi-mnemopi": "15.13.1",
-		"@oh-my-pi/pi-natives": "15.13.1",
-		"@oh-my-pi/pi-tui": "15.13.1",
-		"@oh-my-pi/pi-utils": "15.13.1",
-		"@oh-my-pi/pi-wire": "15.13.1",
-		"@oh-my-pi/snapcompact": "15.13.1",
+		"@oh-my-pi/hashline": "15.13.3",
+		"@oh-my-pi/omp-stats": "15.13.3",
+		"@oh-my-pi/pi-agent-core": "15.13.3",
+		"@oh-my-pi/pi-ai": "15.13.3",
+		"@oh-my-pi/pi-catalog": "15.13.3",
+		"@oh-my-pi/pi-mnemopi": "15.13.3",
+		"@oh-my-pi/pi-natives": "15.13.3",
+		"@oh-my-pi/pi-tui": "15.13.3",
+		"@oh-my-pi/pi-utils": "15.13.3",
+		"@oh-my-pi/pi-wire": "15.13.3",
+		"@oh-my-pi/snapcompact": "15.13.3",
 		"@opentelemetry/api": "^1.9.1",
 		"@opentelemetry/context-async-hooks": "^2.7.1",
 		"@opentelemetry/exporter-trace-otlp-proto": "^0.218.0",

package/src/cli.ts

@@ -14,6 +14,7 @@ try {
  * CLI entry point — registers all commands explicitly and delegates to the
  * lightweight CLI runner from pi-utils.
  */
+import { parentPort } from "node:worker_threads";
 import type { CliConfig } from "@oh-my-pi/pi-utils/cli";
 import {
 	APP_NAME,
@@ -23,7 +24,7 @@ import {
 	setProfile,
 	VERSION,
 } from "@oh-my-pi/pi-utils/dirs";
-import { declareWorkerHostEntry } from "@oh-my-pi/pi-utils/worker-host";
+import { declareWorkerHostEntry, installWorkerInbox } from "@oh-my-pi/pi-utils/worker-host";
 import { installProfileAlias, resolveProfileAliasCommandFromProcess } from "./cli/profile-alias";
 import { extractProfileFlags } from "./cli/profile-bootstrap";
 
@@ -67,6 +68,7 @@ async function runSmokeTest(): Promise<void> {
 	const { smokeTestTinyTitleWorker } = await import("./tiny/title-client");
 	const { smokeTestSttWorker } = await import("./stt/asr-client");
 	const { smokeTestTtsWorker } = await import("./tts/tts-client");
+	const { smokeTestJsEvalWorker } = await import("./eval/js/context-manager");
 	await smokeTestSyncWorker();
 
 	const statsServer = await startServer(0);
@@ -83,18 +85,23 @@ async function runSmokeTest(): Promise<void> {
 
 	await smokeTestTinyTitleWorker();
 	await smokeTestSttWorker();
+	await smokeTestJsEvalWorker();
 	await smokeTestTtsWorker();
 	process.stdout.write("smoke-test: ok\n");
 }
 
-const TINY_WORKER_ARGS = new Set(["--tiny-worker", "__tiny_worker"]);
-const STATS_SYNC_WORKER_ARG = "__omp_stats_sync_worker";
-const TAB_WORKER_ARG = "__omp_tab_worker";
-const JS_EVAL_WORKER_ARG = "__omp_js_eval_worker";
-const STT_WORKER_ARG = "__omp_stt_worker";
-const TTS_WORKER_ARG = "__omp_tts_worker";
+const TINY_WORKER_ARG = "__omp_worker_tiny_inference";
+const STATS_SYNC_WORKER_ARG = "__omp_worker_stats_sync";
+const TAB_WORKER_ARG = "__omp_worker_tab";
+const JS_EVAL_WORKER_ARG = "__omp_worker_js_eval";
+const STT_WORKER_ARG = "__omp_worker_stt";
+const TTS_WORKER_ARG = "__omp_worker_tts";
 
 async function runWorkerEntrypoint(arg: string | undefined): Promise<boolean> {
+	if (arg === TINY_WORKER_ARG) {
+		await runTinyWorker();
+		return true;
+	}
 	if (arg === STATS_SYNC_WORKER_ARG) {
 		// The sync worker handles messages via `self.onmessage`, assigned during
 		// this *async* dynamic import. Bun flushes the worker's initial message
@@ -117,11 +124,20 @@ async function runWorkerEntrypoint(arg: string | undefined): Promise<boolean> {
 		}
 		return true;
 	}
+	// Bun flushes messages the parent posted before spawn once this entry's
+	// top-level evaluation completes, delivering them only to listeners present
+	// at that moment. These worker modules are imported dynamically below, so
+	// their own `parentPort.on("message")` lands after the flush and the parent's
+	// synchronous `init` is dropped. Install a buffering inbox synchronously here
+	// (still inside the entry's sync prefix) so the handshake survives; the worker
+	// module binds the real handler once loaded.
 	if (arg === TAB_WORKER_ARG) {
+		if (parentPort) installWorkerInbox(parentPort);
 		await import("./tools/browser/tab-worker-entry");
 		return true;
 	}
 	if (arg === JS_EVAL_WORKER_ARG) {
+		if (parentPort) installWorkerInbox(parentPort);
 		await import("./eval/js/worker-entry");
 		return true;
 	}
@@ -251,11 +267,8 @@ export async function runCli(argv: string[]): Promise<void> {
 	// synchronous prefix of `runWorkerEntrypoint`, and Bun flushes the
 	// worker's parked initial messages as soon as the entry module's
 	// top-level evaluation finishes.
-	if (TINY_WORKER_ARGS.has(resolvedArgv[0] ?? "")) {
-		await runTinyWorker();
-		return;
-	}
-	if (await runWorkerEntrypoint(resolvedArgv[0])) {
+	if (resolvedArgv[0]?.startsWith("__omp_worker_")) {
+		await runWorkerEntrypoint(resolvedArgv[0]);
 		return;
 	}
 

package/src/config/model-registry.ts

@@ -59,7 +59,7 @@ import {
 	resolveCanonicalVariant,
 	resolveModelReference,
 } from "@oh-my-pi/pi-catalog/identity";
-import { isRecord, logger } from "@oh-my-pi/pi-utils";
+import { isBunTestRuntime, isRecord, logger } from "@oh-my-pi/pi-utils";
 import { parseModelString, resolveProviderModelReference } from "../config/model-resolver";
 import type { AuthStorage, OAuthCredential } from "../session/auth-storage";
 import { type ApiKeyResolverModel, type ApiKeyResolverOptions, createApiKeyResolver } from "./api-key-resolver";
@@ -130,11 +130,13 @@ export function mergeDiscoveredModel<TApi extends Api>(
 	providerOverride?: Pick<ProviderOverride, "baseUrl" | "headers" | "transport">,
 ): Model<TApi> {
 	if (existing) {
+		const supportsTools = model.supportsTools ?? existing.supportsTools;
 		return buildModel({
 			...model,
 			baseUrl: providerOverride?.baseUrl ?? model.baseUrl ?? existing.baseUrl,
 			headers: existing.headers ? { ...existing.headers, ...model.headers } : model.headers,
 			transport: providerOverride?.transport ?? existing.transport ?? model.transport,
+			...(supportsTools !== undefined ? { supportsTools } : {}),
 			compat: model.compatConfig,
 		} as ModelSpec<TApi>);
 	}
@@ -370,6 +372,7 @@ interface ModelPatch {
 	reasoning?: boolean;
 	thinking?: ThinkingConfig;
 	input?: ("text" | "image")[];
+	supportsTools?: boolean;
 	cost?: Partial<Model<Api>["cost"]>;
 	contextWindow?: number;
 	maxTokens?: number;
@@ -395,6 +398,7 @@ function applyModelPatch(base: Model<Api>, patch: ModelPatch, transport: ModelTr
 	if (patch.reasoning !== undefined) result.reasoning = patch.reasoning;
 	if (patch.thinking !== undefined) result.thinking = patch.thinking;
 	if (patch.input !== undefined) result.input = patch.input;
+	if (patch.supportsTools !== undefined) result.supportsTools = patch.supportsTools;
 	if (patch.contextWindow !== undefined) result.contextWindow = patch.contextWindow;
 	if (patch.maxTokens !== undefined) result.maxTokens = patch.maxTokens;
 	if (patch.omitMaxOutputTokens !== undefined) result.omitMaxOutputTokens = patch.omitMaxOutputTokens;
@@ -506,6 +510,7 @@ function buildCustomModelOverlay(
 		reasoning: modelDef.reasoning,
 		thinking: modelDef.thinking,
 		input: modelDef.input,
+		supportsTools: modelDef.supportsTools,
 		cost: modelDef.cost,
 		contextWindow: modelDef.contextWindow,
 		maxTokens: modelDef.maxTokens,
@@ -535,6 +540,7 @@ function finalizeCustomModel(model: CustomModelOverlay, options: CustomModelBuil
 		reference?.cost ??
 		(options.useDefaults ? { input: 0, output: 0, cacheRead: 0, cacheWrite: 0 } : undefined);
 	const input = resolvedModel.input ?? reference?.input ?? (options.useDefaults ? ["text"] : undefined);
+	const supportsTools = resolvedModel.supportsTools ?? reference?.supportsTools;
 	return buildModel({
 		id: resolvedModel.id,
 		name: resolvedModel.name ?? (options.useDefaults ? resolvedModel.id : undefined),
@@ -544,6 +550,7 @@ function finalizeCustomModel(model: CustomModelOverlay, options: CustomModelBuil
 		reasoning: resolvedModel.reasoning ?? reference?.reasoning ?? (options.useDefaults ? false : undefined),
 		thinking: resolvedModel.thinking ?? reference?.thinking,
 		input: input as ("text" | "image")[],
+		...(supportsTools !== undefined ? { supportsTools } : {}),
 		cost,
 		contextWindow: resolvedModel.contextWindow ?? reference?.contextWindow ?? (options.useDefaults ? 128000 : null),
 		maxTokens: resolvedModel.maxTokens ?? reference?.maxTokens ?? (options.useDefaults ? 16384 : null),
@@ -683,7 +690,11 @@ export class ModelRegistry {
 		modelsPath?: string,
 		options?: { fetch?: FetchImpl },
 	) {
-		this.#fetch = options?.fetch ?? fetch;
+		this.#fetch =
+			options?.fetch ??
+			(isBunTestRuntime()
+				? () => Promise.reject(new Error("network disabled in model-registry runtime test"))
+				: fetch);
 		this.#modelsConfigFile = ModelsConfigFile.relocate(modelsPath);
 		this.#cacheDbPath = modelsPath ? path.join(path.dirname(modelsPath), "models.db") : undefined;
 		// Set up fallback resolver for custom provider API keys
@@ -878,10 +889,12 @@ export class ModelRegistry {
 	#mergeResolvedModels(baseModels: Model<Api>[], replacementModels: Model<Api>[]): Model<Api>[] {
 		return mergeByModelKey(baseModels, replacementModels, (existing, replacementModel) => {
 			if (!existing) return replacementModel;
+			const supportsTools = replacementModel.supportsTools ?? existing.supportsTools;
 			return {
 				...replacementModel,
 				contextWindow: replacementModel.contextWindow ?? existing.contextWindow,
 				maxTokens: replacementModel.maxTokens ?? existing.maxTokens,
+				...(supportsTools !== undefined ? { supportsTools } : {}),
 			};
 		});
 	}
@@ -2205,6 +2218,7 @@ export interface ProviderConfigInput {
 		reasoning: boolean;
 		thinking?: ThinkingConfig;
 		input: ("text" | "image")[];
+		supportsTools?: boolean;
 		cost: { input: number; output: number; cacheRead: number; cacheWrite: number };
 		contextWindow: number;
 		maxTokens: number;

package/src/config/models-config-schema.ts

@@ -133,6 +133,7 @@ const ModelDefinitionSchema = z.object({
 	reasoning: z.boolean().optional(),
 	thinking: ModelThinkingSchema.optional(),
 	input: z.array(z.enum(["text", "image"])).optional(),
+	supportsTools: z.boolean().optional(),
 	cost: z
 		.object({
 			input: z.number(),
@@ -155,6 +156,7 @@ export const ModelOverrideSchema = z.object({
 	reasoning: z.boolean().optional(),
 	thinking: ModelThinkingSchema.optional(),
 	input: z.array(z.enum(["text", "image"])).optional(),
+	supportsTools: z.boolean().optional(),
 	cost: z
 		.object({
 			input: z.number().optional(),

package/src/config/models-config.ts

@@ -17,6 +17,7 @@ export interface ProviderValidationModel {
 	id: string;
 	api?: Api;
 	contextWindow?: number;
+	supportsTools?: boolean;
 	maxTokens?: number;
 }
 

package/src/config/settings-schema.ts

@@ -116,6 +116,7 @@ export const TAB_GROUPS: Record<SettingTab, readonly string[]> = {
 		"Magic Keywords",
 		"Startup & Updates",
 		"Power (macOS)",
+		"Agent",
 	],
 	context: ["General", "Compaction", "Rules (TTSR)", "Experimental"],
 	memory: ["General", "Auto-Learn", "Mnemopi", "Hindsight"],
@@ -1719,6 +1720,48 @@ export const SETTINGS_SCHEMA = {
 		},
 	},
 
+	"tools.format": {
+		type: "enum",
+		values: [
+			"auto",
+			"native",
+			"glm",
+			"hermes",
+			"kimi",
+			"xml",
+			"anthropic",
+			"deepseek",
+			"harmony",
+			"pi",
+			"qwen3",
+		] as const,
+		default: "auto",
+		ui: {
+			tab: "context",
+			group: "Experimental",
+			label: "Tool Call Format",
+			description:
+				"Controls how tools are exposed to the model. Auto uses native tool calls unless the selected model is marked as not supporting tools, then falls back to GLM-style in-band tool calls. Native forces provider-native tools; the other values force the named in-band syntax. Applies on session start.",
+			options: [
+				{
+					value: "auto",
+					label: "Auto",
+					description: "Use native tool calls unless the model is known not to support them.",
+				},
+				{ value: "native", label: "Native", description: "Use provider-native tool calls." },
+				{ value: "glm", label: "GLM", description: "Use GLM-style in-band tool calls." },
+				{ value: "hermes", label: "Hermes", description: "Use Hermes-style in-band tool calls." },
+				{ value: "kimi", label: "Kimi", description: "Use Kimi-style in-band tool calls." },
+				{ value: "xml", label: "XML", description: "Use generic XML in-band tool calls." },
+				{ value: "anthropic", label: "Anthropic", description: "Use Anthropic-style in-band tool calls." },
+				{ value: "deepseek", label: "DeepSeek", description: "Use DeepSeek-style in-band tool calls." },
+				{ value: "harmony", label: "Harmony", description: "Use Harmony-style in-band tool calls." },
+				{ value: "pi", label: "Pi", description: "Use Pi-style in-band tool calls." },
+				{ value: "qwen3", label: "Qwen3", description: "Use Qwen3-style in-band tool calls." },
+			],
+		},
+	},
+
 	"snapcompact.shape": {
 		type: "enum",
 		values: ["auto", ...SHAPE_VARIANT_NAMES] as const,
@@ -3183,6 +3226,17 @@ export const SETTINGS_SCHEMA = {
 			description: "Ask the agent to describe the intent of each tool call before executing it",
 		},
 	},
+	"tools.abortOnFabricatedResult": {
+		type: "boolean",
+		default: true,
+		ui: {
+			tab: "tools",
+			group: "Execution",
+			label: "Abort On Fabricated Tool Result",
+			description:
+				"With in-band tool calls, stop the model immediately when it starts hallucinating a tool result mid-turn. Disable to let the model finish generating and discard the fabricated continuation instead.",
+		},
+	},
 
 	"tools.maxTimeout": {
 		type: "number",
@@ -3940,6 +3994,30 @@ export const SETTINGS_SCHEMA = {
 			options: AUTO_THINKING_MODEL_OPTIONS,
 		},
 	},
+	"features.unexpectedStopDetection": {
+		type: "boolean",
+		default: false,
+		ui: {
+			tab: "interaction",
+			group: "Agent",
+			label: "Detect unexpected stops",
+			description:
+				"Use a small model to detect when the assistant says it will continue but stops without tool calls; automatically prompt it to continue.",
+		},
+	},
+	"providers.unexpectedStopModel": {
+		type: "enum",
+		values: TINY_MEMORY_MODEL_VALUES,
+		default: ONLINE_MEMORY_MODEL_KEY,
+		ui: {
+			tab: "providers",
+			group: "Tiny Model",
+			label: "Unexpected Stop Model",
+			description: "Classifier for unexpected-stop detection: online smol by default, or a local on-device model.",
+			condition: "unexpectedStopDetection",
+			options: TINY_MEMORY_MODEL_OPTIONS,
+		},
+	},
 
 	"providers.kimiApiFormat": {
 		type: "enum",

package/src/edit/hashline/block-resolver.ts

@@ -1,5 +1,5 @@
 /**
- * Tree-sitter-backed {@link BlockResolver} for the hashline `replace block N:`
+ * Tree-sitter-backed {@link BlockResolver} for the hashline block replace
  * operator. Bridges the pure hashline seam to the native `blockRangeAt`
  * primitive in `@oh-my-pi/pi-natives`, which infers the language from the file
  * path and returns the 1-indexed line span of the syntactic block beginning on

package/src/edit/hashline/execute.ts

@@ -98,12 +98,7 @@ interface RenderedSection {
 }
 
 function formatBlockResolution(resolution: BlockResolution): string {
-	const op =
-		resolution.op === "delete"
-			? "delete block"
-			: resolution.op === "insert_after"
-				? "insert after block"
-				: "replace block";
+	const op = resolution.op === "delete" ? "DEL.BLK" : resolution.op === "insert_after" ? "INS.BLK.POST" : "SWAP.BLK";
 	const lines = resolution.end - resolution.start + 1;
 	const span =
 		resolution.start === resolution.end ? `line ${resolution.start}` : `lines ${resolution.start}-${resolution.end}`;

package/src/edit/index.ts

@@ -2,7 +2,9 @@ import { MismatchError as HashlineMismatchError } from "@oh-my-pi/hashline";
 import hashlineGrammar from "@oh-my-pi/hashline/grammar.lark" with { type: "text" };
 import hashlineDescription from "@oh-my-pi/hashline/prompt.md" with { type: "text" };
 import type { AgentTool, AgentToolContext, AgentToolResult, AgentToolUpdateCallback } from "@oh-my-pi/pi-agent-core";
+import type { ToolExample } from "@oh-my-pi/pi-ai";
 import { prompt } from "@oh-my-pi/pi-utils";
+import type { z } from "zod/v4";
 import {
 	createLspWritethrough,
 	type FileDiagnosticsResult,
@@ -50,6 +52,7 @@ type EditParams = ReplaceParams | PatchParams | HashlineParams | ApplyPatchParam
 type EditModeDefinition = {
 	description: (session: ToolSession) => string;
 	parameters: TInput;
+	examples?: readonly ToolExample[];
 	execute: (
 		tool: EditTool,
 		params: EditParams,
@@ -356,6 +359,10 @@ export class EditTool implements AgentTool<TInput> {
 		return this.#getModeDefinition().parameters;
 	}
 
+	get examples(): readonly ToolExample[] | undefined {
+		return this.#getModeDefinition().examples;
+	}
+
 	/**
 	 * When in `apply_patch` mode, expose the Codex Lark grammar so providers
 	 * that support OpenAI-style custom tools can emit a grammar-constrained
@@ -404,6 +411,39 @@ export class EditTool implements AgentTool<TInput> {
 			patch: {
 				description: () => prompt.render(patchDescription),
 				parameters: patchEditSchema,
+				examples: [
+					{
+						caption: "Create",
+						call: { path: "hello.txt", edits: [{ op: "create", diff: "Hello\n" }] },
+					},
+					{
+						caption: "Update",
+						call: {
+							path: "src/app.py",
+							edits: [
+								{
+									op: "update",
+									diff: "@@ def greet():\n def greet():\n-print('Hi')\n+print('Hello')\n",
+								},
+							],
+						},
+					},
+					{
+						caption: "Rename",
+						call: {
+							path: "src/app.py",
+							edits: [{ op: "update", rename: "src/main.py", diff: "@@\n …\n" }],
+						},
+					},
+					{
+						caption: "Delete",
+						call: { path: "obsolete.txt", edits: [{ op: "delete" }] },
+					},
+					{
+						caption: "Multiple entries",
+						note: "All entries in one call apply to the top-level `path`; use separate calls for different files.",
+					},
+				] satisfies readonly ToolExample<z.input<typeof patchEditSchema>>[],
 				execute: (
 					tool: EditTool,
 					params: EditParams,
@@ -432,6 +472,14 @@ export class EditTool implements AgentTool<TInput> {
 			apply_patch: {
 				description: () => prompt.render(applyPatchDescription),
 				parameters: applyPatchSchema,
+				examples: [
+					{
+						caption: "Apply a combined patch file",
+						call: {
+							input: '*** Begin Patch\n*** Add File: hello.txt\n+Hello world\n*** Update File: src/app.py\n*** Move to: src/main.py\n@@ def greet():\n-print("Hi")\n+print("Hello, world!")\n*** Delete File: obsolete.txt\n*** End Patch\n',
+						},
+					},
+				] satisfies readonly ToolExample<z.input<typeof applyPatchSchema>>[],
 				execute: (
 					tool: EditTool,
 					params: EditParams,