feat(agent): AgentHarness streams through a required Models instance (phase 6)

AgentHarnessOptions.models is required; the harness stream path,
compaction, and branch summarization go through models.streamSimple()/
completeSimple() instead of the compat globals. getApiKeyAndHeaders
stays and wins per-field over provider-resolved auth, but is no longer
required: without it, requests resolve through provider auth.

compact()/generateSummary()/generateBranchSummary() take a Models
parameter; explicit apiKey becomes optional. StreamFn is redefined
structurally (Models.streamSimple satisfies it), dropping the compat
type dependency from agent types.

Harness tests build per-file Models collections with fauxProvider()
and unique provider ids instead of mutating the global api-registry.

packages/agent/src/harness/compaction/branch-summarization.ts

@@ -1,5 +1,5 @@
-import type { Model } from "@earendil-works/pi-ai/compat";
-import { completeSimple } from "@earendil-works/pi-ai/compat";
+import type { Model, Models } from "@earendil-works/pi-ai";
+
 import type { AgentMessage } from "../../types.ts";
 import {
 	convertToLlm,
@@ -49,10 +49,12 @@ export interface CollectEntriesResult {
 
 /** Options for generating a branch summary. */
 export interface GenerateBranchSummaryOptions {
+	/** Provider collection the summarization request goes through. */
+	models: Models;
 	/** Model used for summarization. */
 	model: Model<any>;
-	/** API key forwarded to the provider. */
-	apiKey: string;
+	/** Explicit API key; wins over provider-resolved auth. */
+	apiKey?: string;
 	/** Optional request headers forwarded to the provider. */
 	headers?: Record<string, string>;
 	/** Abort signal for the summarization request. */
@@ -202,7 +204,16 @@ export async function generateBranchSummary(
 	entries: SessionTreeEntry[],
 	options: GenerateBranchSummaryOptions,
 ): Promise<Result<BranchSummaryResult, BranchSummaryError>> {
-	const { model, apiKey, headers, signal, customInstructions, replaceInstructions, reserveTokens = 16384 } = options;
+	const {
+		models,
+		model,
+		apiKey,
+		headers,
+		signal,
+		customInstructions,
+		replaceInstructions,
+		reserveTokens = 16384,
+	} = options;
 	const contextWindow = model.contextWindow || 128000;
 	const tokenBudget = contextWindow - reserveTokens;
 
@@ -230,7 +241,7 @@ export async function generateBranchSummary(
 			timestamp: Date.now(),
 		},
 	];
-	const response = await completeSimple(
+	const response = await models.completeSimple(
 		model,
 		{ systemPrompt: SUMMARIZATION_SYSTEM_PROMPT, messages: summarizationMessages },
 		{ apiKey, headers, signal, maxTokens: 2048 },

packages/agent/src/harness/compaction/compaction.ts

@@ -1,5 +1,4 @@
-import type { AssistantMessage, ImageContent, Model, TextContent, Usage } from "@earendil-works/pi-ai/compat";
-import { completeSimple } from "@earendil-works/pi-ai/compat";
+import type { AssistantMessage, ImageContent, Model, Models, TextContent, Usage } from "@earendil-works/pi-ai";
 import type { AgentMessage, ThinkingLevel } from "../../types.ts";
 import {
 	convertToLlm,
@@ -455,9 +454,10 @@ Keep each section concise. Preserve exact file paths, function names, and error
 /** Generate or update a conversation summary for compaction. */
 export async function generateSummary(
 	currentMessages: AgentMessage[],
+	models: Models,
 	model: Model<any>,
 	reserveTokens: number,
-	apiKey: string,
+	apiKey?: string,
 	headers?: Record<string, string>,
 	signal?: AbortSignal,
 	customInstructions?: string,
@@ -493,7 +493,7 @@ export async function generateSummary(
 			? { maxTokens, signal, apiKey, headers, reasoning: thinkingLevel }
 			: { maxTokens, signal, apiKey, headers };
 
-	const response = await completeSimple(
+	const response = await models.completeSimple(
 		model,
 		{ systemPrompt: SUMMARIZATION_SYSTEM_PROMPT, messages: summarizationMessages },
 		completionOptions,
@@ -626,8 +626,9 @@ export { serializeConversation } from "./utils.ts";
 /** Generate compaction summary data from prepared session history. */
 export async function compact(
 	preparation: CompactionPreparation,
+	models: Models,
 	model: Model<any>,
-	apiKey: string,
+	apiKey?: string,
 	headers?: Record<string, string>,
 	customInstructions?: string,
 	signal?: AbortSignal,
@@ -655,6 +656,7 @@ export async function compact(
 			messagesToSummarize.length > 0
 				? generateSummary(
 						messagesToSummarize,
+						models,
 						model,
 						settings.reserveTokens,
 						apiKey,
@@ -667,6 +669,7 @@ export async function compact(
 				: Promise.resolve(ok<string, CompactionError>("No prior history.")),
 			generateTurnPrefixSummary(
 				turnPrefixMessages,
+				models,
 				model,
 				settings.reserveTokens,
 				apiKey,
@@ -681,6 +684,7 @@ export async function compact(
 	} else {
 		const summaryResult = await generateSummary(
 			messagesToSummarize,
+			models,
 			model,
 			settings.reserveTokens,
 			apiKey,
@@ -706,9 +710,10 @@ export async function compact(
 }
 async function generateTurnPrefixSummary(
 	messages: AgentMessage[],
+	models: Models,
 	model: Model<any>,
 	reserveTokens: number,
-	apiKey: string,
+	apiKey?: string,
 	headers?: Record<string, string>,
 	signal?: AbortSignal,
 	thinkingLevel?: ThinkingLevel,
@@ -728,7 +733,7 @@ async function generateTurnPrefixSummary(
 		},
 	];
 
-	const response = await completeSimple(
+	const response = await models.completeSimple(
 		model,
 		{ systemPrompt: SUMMARIZATION_SYSTEM_PROMPT, messages: summarizationMessages },
 		model.reasoning && thinkingLevel && thinkingLevel !== "off"