fix: complete extension usage accounting

closes #6509

packages/coding-agent/src/core/compaction/compaction.ts

@@ -581,6 +581,37 @@ export async function generateSummary(
 	thinkingLevel?: ThinkingLevel,
 	streamFn?: StreamFn,
 	env?: Record<string, string>,
+): Promise<string> {
+	return (
+		await generateSummaryWithUsage(
+			currentMessages,
+			model,
+			reserveTokens,
+			apiKey,
+			headers,
+			signal,
+			customInstructions,
+			previousSummary,
+			thinkingLevel,
+			streamFn,
+			env,
+		)
+	).text;
+}
+
+/** Generate or update a conversation summary and return its provider usage. */
+export async function generateSummaryWithUsage(
+	currentMessages: AgentMessage[],
+	model: Model<any>,
+	reserveTokens: number,
+	apiKey: string | undefined,
+	headers?: Record<string, string>,
+	signal?: AbortSignal,
+	customInstructions?: string,
+	previousSummary?: string,
+	thinkingLevel?: ThinkingLevel,
+	streamFn?: StreamFn,
+	env?: Record<string, string>,
 ): Promise<{ text: string; usage: Usage }> {
 	const maxTokens = Math.min(
 		Math.floor(0.8 * reserveTokens),
@@ -790,7 +821,7 @@ export async function compact(
 		let historyText = "No prior history.";
 		let historyUsage: Usage | undefined;
 		if (messagesToSummarize.length > 0) {
-			const historyResult = await generateSummary(
+			const historyResult = await generateSummaryWithUsage(
 				messagesToSummarize,
 				model,
 				settings.reserveTokens,
@@ -822,7 +853,7 @@ export async function compact(
 		summaryUsage = historyUsage ? combineUsage(historyUsage, turnPrefixResult.usage) : turnPrefixResult.usage;
 	} else {
 		// Just generate history summary
-		const result = await generateSummary(
+		const result = await generateSummaryWithUsage(
 			messagesToSummarize,
 			model,
 			settings.reserveTokens,

packages/coding-agent/src/core/usage-totals.ts

@@ -1,4 +1,5 @@
 import type { Usage } from "@earendil-works/pi-ai/compat";
+import type { SessionEntry } from "./session-manager.ts";
 
 export interface UsageTotals {
 	input: number;
@@ -25,3 +26,45 @@ export function addUsageToTotals(totals: UsageTotals, usage: Usage): void {
 	totals.cacheWrite += usage.cacheWrite;
 	totals.cost += usage.cost.total;
 }
+
+export interface UsageCostBreakdownEntry {
+	key: string;
+	cost: number;
+	tokens: number;
+}
+
+/** Group attributable assistant usage by model and all other usage into a separate bucket. */
+export function getUsageCostBreakdown(entries: SessionEntry[]): UsageCostBreakdownEntry[] {
+	const totalsByKey = new Map<string, UsageTotals>();
+
+	for (const entry of entries) {
+		let key: string | undefined;
+		let usage: Usage | undefined;
+		if (entry.type === "message" && entry.message.role === "assistant") {
+			key = `${entry.message.provider}/${entry.message.responseModel ?? entry.message.model}`;
+			usage = entry.message.usage;
+		} else if (entry.type === "message" && entry.message.role === "toolResult" && entry.message.usage) {
+			key = "Tools/summaries";
+			usage = entry.message.usage;
+		} else if ((entry.type === "branch_summary" || entry.type === "compaction") && entry.usage) {
+			key = "Tools/summaries";
+			usage = entry.usage;
+		}
+		if (!key || !usage) continue;
+
+		let totals = totalsByKey.get(key);
+		if (!totals) {
+			totals = createUsageTotals();
+			totalsByKey.set(key, totals);
+		}
+		addUsageToTotals(totals, usage);
+	}
+
+	return Array.from(totalsByKey, ([key, totals]) => ({
+		key,
+		cost: totals.cost,
+		tokens: totals.input + totals.output + totals.cacheRead + totals.cacheWrite,
+	}))
+		.filter((entry) => entry.cost > 0 || entry.tokens > 0)
+		.sort((a, b) => b.cost - a.cost);
+}

packages/coding-agent/src/index.ts

@@ -42,6 +42,7 @@ export {
 	type GenerateBranchSummaryOptions,
 	generateBranchSummary,
 	generateSummary,
+	generateSummaryWithUsage,
 	getLastAssistantUsage,
 	prepareBranchEntries,
 	serializeConversation,

packages/coding-agent/src/modes/interactive/interactive-mode.ts

@@ -86,6 +86,7 @@ import type { SourceInfo } from "../../core/source-info.ts";
 import { isInstallTelemetryEnabled } from "../../core/telemetry.ts";
 import type { TruncationResult } from "../../core/tools/truncate.ts";
 import { hasTrustRequiringProjectResources, ProjectTrustStore } from "../../core/trust-manager.ts";
+import { getUsageCostBreakdown } from "../../core/usage-totals.ts";
 import { getChangelogPath, getNewEntries, normalizeChangelogLinks, parseChangelog } from "../../utils/changelog.ts";
 import { copyToClipboard, readClipboardText } from "../../utils/clipboard.ts";
 import { extensionForImageMimeType, readClipboardImage } from "../../utils/clipboard-image.ts";
@@ -5597,22 +5598,9 @@ export class InteractiveMode {
 		const cacheWaste = computeCacheWaste(entries, this.session.modelRuntime);
 
 		// Cost/token totals per provider/model actually used (e.g. OpenRouter `auto`
-		// resolves to a concrete responseModel), sorted by cost descending.
-		const perModelMap = new Map<string, { key: string; cost: number; tokens: number }>();
-		for (const entry of entries) {
-			if (entry.type !== "message" || entry.message.role !== "assistant") continue;
-			const message = entry.message;
-			const usage = message.usage;
-			const key = `${message.provider}/${message.responseModel ?? message.model}`;
-			let bucket = perModelMap.get(key);
-			if (!bucket) {
-				bucket = { key, cost: 0, tokens: 0 };
-				perModelMap.set(key, bucket);
-			}
-			bucket.cost += usage.cost.total;
-			bucket.tokens += usage.input + usage.output + usage.cacheRead + usage.cacheWrite;
-		}
-		const perModel = Array.from(perModelMap.values()).sort((a, b) => b.cost - a.cost);
+		// resolves to a concrete responseModel). Usage without model attribution is
+		// grouped separately so the breakdown reconciles with the session total.
+		const usageBreakdown = getUsageCostBreakdown(entries);
 
 		let info = `${theme.bold("Session Info")}\n\n`;
 		if (sessionName) {
@@ -5646,8 +5634,8 @@ export class InteractiveMode {
 		if (stats.cost > 0 || cacheWaste.missedTokens > 0) {
 			info += `\n${theme.bold("Cost")}\n`;
 			info += `${theme.fg("dim", "Total:")} $${stats.cost.toFixed(3)}`;
-			if (perModel.length > 1) {
-				for (const entry of perModel) {
+			if (usageBreakdown.length > 1) {
+				for (const entry of usageBreakdown) {
 					info += `\n  ${theme.fg("dim", `${entry.key}:`)} $${entry.cost.toFixed(3)} ${theme.fg("dim", `(${formatTokens(entry.tokens)} tokens)`)}`;
 				}
 			}