@oh-my-pi/pi-coding-agent 15.5.13 → 15.6.0

package/src/task/render.ts

@@ -10,6 +10,7 @@ import { Container, Text } from "@oh-my-pi/pi-tui";
 import { formatNumber } from "@oh-my-pi/pi-utils";
 import { settings } from "../config/settings";
 import type { RenderResultOptions } from "../extensibility/custom-tools/types";
+import { formatContextUsage } from "../modes/components/status-line/context-thresholds";
 import type { Theme } from "../modes/theme/theme";
 import {
 	formatBadge,
@@ -29,7 +30,7 @@ import {
 } from "../tools/review";
 import { Ellipsis, Hasher, type RenderCache, renderStatusLine } from "../tui";
 import { subprocessToolRegistry } from "./subprocess-tool-registry";
-import type { AgentProgress, SingleResult, TaskParams, TaskToolDetails } from "./types";
+import type { AgentProgress, SingleResult, TaskItem, TaskParams, TaskToolDetails } from "./types";
 
 /**
  * Get status icon for agent state.
@@ -51,7 +52,9 @@ function getStatusIcon(status: AgentProgress["status"], theme: Theme, spinnerFra
 	}
 }
 
-/** Append tool-count, context, cumulative-tokens, and cost stats to a status line string. */
+/**
+ * Append tool-count, context, and cost stats to a status line string.
+ */
 function appendAgentStats(
 	line: string,
 	opts: {
@@ -66,22 +69,15 @@ function appendAgentStats(
 	theme: Theme,
 ): string {
 	if (opts.toolCount) {
-		line += `${theme.sep.dot}${theme.fg("dim", `${opts.toolCount} tools`)}`;
+		line += `${theme.sep.dot}${theme.fg("dim", `${formatNumber(opts.toolCount)} ${theme.icon.extensionTool}`)}`;
 	}
-	// Current per-turn context — what the user reads as "how full is the context".
-	// Cumulative tokens (billing volume) renders separately with a Σ sigil to avoid
-	// being mistaken for current window pressure.
+	// Current per-turn context — match the status line's `<pct>%/<window>` gauge (e.g. `5.1%/1M`).
 	if (opts.contextTokens && opts.contextTokens > 0) {
 		const ctx =
 			opts.contextWindow && opts.contextWindow > 0
-				? `${formatNumber(opts.contextTokens)}/${formatNumber(opts.contextWindow)} ctx`
-				: `${formatNumber(opts.contextTokens)} ctx`;
+				? formatContextUsage((opts.contextTokens / opts.contextWindow) * 100, opts.contextWindow)
+				: `${formatNumber(opts.contextTokens)}`;
 		line += `${theme.sep.dot}${theme.fg("dim", ctx)}`;
-		if (opts.tokens > 0) {
-			line += `${theme.sep.dot}${theme.fg("dim", `Σ${formatNumber(opts.tokens)}`)}`;
-		}
-	} else if (opts.tokens > 0) {
-		line += `${theme.sep.dot}${theme.fg("dim", `Σ${formatNumber(opts.tokens)}`)}`;
 	}
 	if (opts.cost > 0) {
 		line += `${theme.sep.dot}${theme.fg("statusLineCost", `$${opts.cost.toFixed(2)}`)}`;
@@ -490,20 +486,63 @@ function formatOutputInline(data: unknown, theme: Theme, maxWidth = 80): string
 	return `Output: ${pairs.join(", ")}`;
 }
 
+/**
+ * Render the per-task list (`id` + ui `description`) for the streaming call
+ * preview. The args stream in token by token, so the array grows over time and
+ * trailing entries may be partially parsed — every field access is defensive.
+ */
+function renderTaskItemLines(
+	tasks: TaskItem[] | undefined,
+	contPrefix: string,
+	expanded: boolean,
+	theme: Theme,
+): string[] {
+	const items = tasks ?? [];
+	if (items.length === 0) return [];
+
+	const branch = theme.fg("dim", theme.tree.branch);
+	const last = theme.fg("dim", theme.tree.last);
+	const cap = expanded ? items.length : Math.min(items.length, 12);
+	const truncated = cap < items.length;
+
+	const lines: string[] = [];
+	for (let i = 0; i < cap; i++) {
+		const task = items[i] as Partial<TaskItem> | undefined;
+		const isLastLine = !truncated && i === items.length - 1;
+		const connector = isLastLine ? last : branch;
+		const rawId = task?.id?.trim();
+		const idLabel = rawId ? formatTaskId(rawId) : `#${i + 1}`;
+		let line = `${contPrefix}${connector} ${theme.fg("accent", theme.bold(idLabel))}`;
+		const desc = task?.description?.trim();
+		if (desc) {
+			line += `: ${theme.fg("muted", truncateToWidth(replaceTabs(desc), 64))}`;
+		}
+		lines.push(line);
+	}
+	if (truncated) {
+		lines.push(`${contPrefix}${last} ${theme.fg("dim", formatMoreItems(items.length - cap, "agent"))}`);
+	}
+	return lines;
+}
+
 /**
  * Render the tool call arguments.
  */
-export function renderCall(args: TaskParams, _options: RenderResultOptions, theme: Theme): Component {
+export function renderCall(
+	args: TaskParams,
+	options: RenderResultOptions & { renderContext?: { hasResult?: boolean } },
+	theme: Theme,
+): Component {
 	const lines: string[] = [];
 	lines.push(renderStatusLine({ icon: "pending", title: "Task", description: args.agent }, theme));
 
-	const contextTemplate = args.context ?? "";
-	const context = contextTemplate.trim();
+	const context = (args.context ?? "").trim();
 	const hasContext = context.length > 0;
 	const branch = theme.fg("dim", theme.tree.branch);
 	const last = theme.fg("dim", theme.tree.last);
 	const vertical = theme.fg("dim", theme.tree.vertical);
 	const showIsolated = "isolated" in args && args.isolated === true;
+	const taskCount = args.tasks?.length ?? 0;
 
 	if (hasContext) {
 		lines.push(` ${branch} ${theme.fg("dim", "Context")}`);
@@ -511,19 +550,23 @@ export function renderCall(args: TaskParams, _options: RenderResultOptions, them
 			const content = line ? theme.fg("muted", replaceTabs(line)) : "";
 			lines.push(` ${vertical}  ${content}`);
 		}
-		const taskPrefix = showIsolated ? branch : last;
-		lines.push(
-			` ${taskPrefix} ${theme.fg("dim", "Tasks")}: ${theme.fg("muted", `${args.tasks?.length ?? 0} agents`)}`,
-		);
-		if (showIsolated) {
-			lines.push(` ${last} ${theme.fg("dim", "Isolated")}: ${theme.fg("muted", "true")}`);
-		}
-		return new Text(lines.join("\n"), 0, 0);
 	}
 
-	lines.push(`${theme.fg("dim", "Tasks")}: ${theme.fg("muted", `${args.tasks?.length ?? 0} agents`)}`);
+	// `Tasks` is the last child unless the isolation flag follows it.
+	const tasksIsLast = !showIsolated;
+	const tasksPrefix = tasksIsLast ? last : branch;
+	lines.push(` ${tasksPrefix} ${theme.fg("dim", "Tasks")} ${theme.fg("muted", `(${taskCount})`)}`);
+	const tasksContPrefix = tasksIsLast ? "    " : ` ${vertical}  `;
+	// The per-task preview list only exists to surface dispatched agents while
+	// the call args stream in. Once a result snapshot exists, `renderResult`
+	// draws the same agents as progress/result lines (id + description), so
+	// emitting the preview here would render every task twice.
+	if (!options.renderContext?.hasResult) {
+		lines.push(...renderTaskItemLines(args.tasks, tasksContPrefix, options.expanded, theme));
+	}
+
 	if (showIsolated) {
-		lines.push(`${theme.fg("dim", "Isolated")}: ${theme.fg("muted", "true")}`);
+		lines.push(` ${last} ${theme.fg("dim", "Isolated")}: ${theme.fg("muted", "true")}`);
 	}
 
 	return new Text(lines.join("\n"), 0, 0);

package/src/tiny/models.ts

@@ -0,0 +1,217 @@
+/** Default session-title model: the online pi/smol path (no local download / CPU inference). */
+export const ONLINE_TINY_TITLE_MODEL_KEY = "online";
+/** Local model the `tiny-models` CLI downloads when none is named. Not the session-title default — that is {@link ONLINE_TINY_TITLE_MODEL_KEY}. */
+export const DEFAULT_TINY_TITLE_LOCAL_MODEL_KEY = "lfm2-700m";
+
+export interface TinyTitleLocalModelSpec {
+	key: string;
+	repo: string;
+	dtype: "q4";
+	label: string;
+	description: string;
+	contextNote: string;
+}
+
+export const TINY_TITLE_LOCAL_MODELS = [
+	{
+		key: "lfm2-350m",
+		repo: "onnx-community/LFM2-350M-ONNX",
+		dtype: "q4",
+		label: "LFM2 350M",
+		description: "Recommended local model; best speed/quality balance, about 212 MB cached.",
+		contextNote: "Best local default from the CPU title-generation spike.",
+	},
+	{
+		key: "qwen3-0.6b",
+		repo: "onnx-community/Qwen3-0.6B-ONNX",
+		dtype: "q4",
+		label: "Qwen3 0.6B",
+		description: "Most robust local option; slower first load, about 500 MB cached.",
+		contextNote: "Use when title quality matters more than local startup cost.",
+	},
+	{
+		key: "gemma-270m",
+		repo: "onnx-community/gemma-3-270m-it-ONNX",
+		dtype: "q4",
+		label: "Gemma 270M",
+		description: "Smallest viable local option; lower quality, lowest cache footprint.",
+		contextNote: "Use on constrained machines that still need local titles.",
+	},
+	{
+		key: "qwen2.5-0.5b",
+		repo: "onnx-community/Qwen2.5-0.5B-Instruct",
+		dtype: "q4",
+		label: "Qwen2.5 0.5B",
+		description: "Balanced local fallback; moderate quality and cache footprint.",
+		contextNote: "Useful when Qwen3 is too heavy but Gemma quality is insufficient.",
+	},
+	{
+		key: "lfm2-700m",
+		repo: "onnx-community/LFM2-700M-ONNX",
+		dtype: "q4",
+		label: "LFM2 700M",
+		description: "Highest-quality local option; larger and slower than LFM2 350M.",
+		contextNote: "Use when local title quality is preferred over startup cost.",
+	},
+] as const satisfies readonly TinyTitleLocalModelSpec[];
+
+export const TINY_TITLE_MODEL_VALUES = [
+	ONLINE_TINY_TITLE_MODEL_KEY,
+	"lfm2-350m",
+	"qwen3-0.6b",
+	"gemma-270m",
+	"qwen2.5-0.5b",
+	"lfm2-700m",
+] as const;
+
+export type TinyTitleModelKey = (typeof TINY_TITLE_MODEL_VALUES)[number];
+export type TinyTitleLocalModelKey = (typeof TINY_TITLE_LOCAL_MODELS)[number]["key"];
+
+type MissingTinyTitleModelValue = Exclude<
+	typeof ONLINE_TINY_TITLE_MODEL_KEY | TinyTitleLocalModelKey,
+	TinyTitleModelKey
+>;
+type ExtraTinyTitleModelValue = Exclude<TinyTitleModelKey, typeof ONLINE_TINY_TITLE_MODEL_KEY | TinyTitleLocalModelKey>;
+const TINY_TITLE_MODEL_VALUES_MATCH_REGISTRY: MissingTinyTitleModelValue extends never
+	? ExtraTinyTitleModelValue extends never
+		? true
+		: never
+	: never = true;
+void TINY_TITLE_MODEL_VALUES_MATCH_REGISTRY;
+
+export const TINY_TITLE_MODEL_OPTIONS = [
+	{
+		value: ONLINE_TINY_TITLE_MODEL_KEY,
+		label: "Online (pi/smol)",
+		description: "Current online title generation path; no local model download or CPU inference.",
+	},
+	...TINY_TITLE_LOCAL_MODELS.map(model => ({
+		value: model.key,
+		label: model.label,
+		description: model.description,
+	})),
+] satisfies ReadonlyArray<{ value: TinyTitleModelKey; label: string; description: string }>;
+
+export function isTinyTitleLocalModelKey(value: string): value is TinyTitleLocalModelKey {
+	return TINY_TITLE_LOCAL_MODELS.some(model => model.key === value);
+}
+
+export function getTinyTitleModelSpec(key: TinyTitleLocalModelKey): (typeof TINY_TITLE_LOCAL_MODELS)[number] {
+	const spec = TINY_TITLE_LOCAL_MODELS.find(model => model.key === key);
+	if (!spec) throw new Error(`Unknown tiny title model: ${key}`);
+	return spec;
+}
+
+/** Default memory model: the online path (the configured smol / remote LLM; no local download). */
+export const ONLINE_MEMORY_MODEL_KEY = "online";
+/** Recommended local model for memory tasks when none is named. */
+export const DEFAULT_MEMORY_LOCAL_MODEL_KEY = "qwen3-1.7b";
+
+/**
+ * Local models for Mnemosyne memory tasks (fact extraction + consolidation).
+ * These are larger (1B-1.7B) than the title models: structured extraction and
+ * faithful summarization need more capacity than 3-6 word titles. All q4, CPU.
+ * Ranking/recipe rationale lives in docs/local-models.md.
+ */
+export const TINY_MEMORY_LOCAL_MODELS = [
+	{
+		key: "qwen3-1.7b",
+		repo: "onnx-community/Qwen3-1.7B-ONNX",
+		dtype: "q4",
+		label: "Qwen3 1.7B",
+		description:
+			"Recommended; most disciplined extraction (ignores chit-chat), good consolidation, about 1.1 GB cached.",
+		contextNote: "Best single-model pick for memory from the CPU experiment.",
+	},
+	{
+		key: "gemma-3-1b",
+		repo: "onnx-community/gemma-3-1b-it-ONNX",
+		dtype: "q4",
+		label: "Gemma 3 1B",
+		description: "Best consolidation/dedup; lighter footprint, but leaks small talk during extraction.",
+		contextNote: "Use when consolidation quality and size matter most.",
+	},
+	{
+		key: "qwen2.5-1.5b",
+		repo: "onnx-community/Qwen2.5-1.5B-Instruct",
+		dtype: "q4",
+		label: "Qwen2.5 1.5B",
+		description: "Best extraction granularity (atomic facts); weaker consolidation.",
+		contextNote: "Use when fine-grained, deduplicatable facts matter more than summaries.",
+	},
+	{
+		key: "lfm2-1.2b",
+		repo: "onnx-community/LFM2-1.2B-ONNX",
+		dtype: "q4",
+		label: "LFM2 1.2B",
+		description: "Fastest load; solid all-rounder, slightly noisier extraction labels.",
+		contextNote: "Use when local startup cost is the priority.",
+	},
+] as const satisfies readonly TinyTitleLocalModelSpec[];
+
+export const TINY_MEMORY_MODEL_VALUES = [
+	ONLINE_MEMORY_MODEL_KEY,
+	"qwen3-1.7b",
+	"gemma-3-1b",
+	"qwen2.5-1.5b",
+	"lfm2-1.2b",
+] as const;
+
+export type TinyMemoryModelKey = (typeof TINY_MEMORY_MODEL_VALUES)[number];
+export type TinyMemoryLocalModelKey = (typeof TINY_MEMORY_LOCAL_MODELS)[number]["key"];
+
+type MissingTinyMemoryModelValue = Exclude<
+	typeof ONLINE_MEMORY_MODEL_KEY | TinyMemoryLocalModelKey,
+	TinyMemoryModelKey
+>;
+type ExtraTinyMemoryModelValue = Exclude<TinyMemoryModelKey, typeof ONLINE_MEMORY_MODEL_KEY | TinyMemoryLocalModelKey>;
+const TINY_MEMORY_MODEL_VALUES_MATCH_REGISTRY: MissingTinyMemoryModelValue extends never
+	? ExtraTinyMemoryModelValue extends never
+		? true
+		: never
+	: never = true;
+void TINY_MEMORY_MODEL_VALUES_MATCH_REGISTRY;
+
+export const TINY_MEMORY_MODEL_OPTIONS = [
+	{
+		value: ONLINE_MEMORY_MODEL_KEY,
+		label: "Online (smol/remote)",
+		description: "Use the configured Mnemosyne LLM mode (smol or remote); no local model download or CPU inference.",
+	},
+	...TINY_MEMORY_LOCAL_MODELS.map(model => ({
+		value: model.key,
+		label: model.label,
+		description: model.description,
+	})),
+] satisfies ReadonlyArray<{ value: TinyMemoryModelKey; label: string; description: string }>;
+
+export function isTinyMemoryLocalModelKey(value: string): value is TinyMemoryLocalModelKey {
+	return TINY_MEMORY_LOCAL_MODELS.some(model => model.key === value);
+}
+
+export function getTinyMemoryModelSpec(key: TinyMemoryLocalModelKey): (typeof TINY_MEMORY_LOCAL_MODELS)[number] {
+	const spec = TINY_MEMORY_LOCAL_MODELS.find(model => model.key === key);
+	if (!spec) throw new Error(`Unknown tiny memory model: ${key}`);
+	return spec;
+}
+
+/** Any local model key (title or memory), used by the shared inference worker. */
+export type TinyLocalModelKey = TinyTitleLocalModelKey | TinyMemoryLocalModelKey;
+
+/** Resolve a local model spec by key across both the title and memory registries. */
+export function getTinyLocalModelSpec(key: string): TinyTitleLocalModelSpec | undefined {
+	return (
+		TINY_TITLE_LOCAL_MODELS.find(model => model.key === key) ??
+		TINY_MEMORY_LOCAL_MODELS.find(model => model.key === key)
+	);
+}
+
+export function isTinyLocalModelKey(value: string): value is TinyLocalModelKey {
+	return getTinyLocalModelSpec(value) !== undefined;
+}
+
+/** Combined local model registry (title + memory) for the shared tiny-models CLI. */
+export const TINY_LOCAL_MODELS = [
+	...TINY_TITLE_LOCAL_MODELS,
+	...TINY_MEMORY_LOCAL_MODELS,
+] as const satisfies readonly TinyTitleLocalModelSpec[];

package/src/tiny/text.ts

@@ -0,0 +1,19 @@
+export const MAX_TITLE_INPUT_CHARS = 2000;
+
+export function truncateTitleInput(message: string): string {
+	return message.length > MAX_TITLE_INPUT_CHARS ? `${message.slice(0, MAX_TITLE_INPUT_CHARS)}…` : message;
+}
+
+export function formatTitleUserMessage(message: string): string {
+	return `<user-message>\n${truncateTitleInput(message)}\n</user-message>`;
+}
+
+export function normalizeGeneratedTitle(value: string | null | undefined): string | null {
+	const firstLine = value?.trim().split(/\r?\n/, 1)[0]?.trim();
+	if (!firstLine) return null;
+	const title = firstLine
+		.replace(/^["']|["']$/g, "")
+		.replace(/[.!?]$/, "")
+		.trim();
+	return title || null;
+}