fix(coding-agent): preserve bundled contextWindow/maxTokens when discovery returns sentinel fallbacks

When cached or freshly-discovered provider models carry UNK_CONTEXT_WINDOW
(222222) / UNK_MAX_TOKENS (8888) sentinels, #mergeResolvedModels was
replacing the bundled model wholesale — wiping out the correct values.

Switch to a field-level merge that preserves the bundled model's
contextWindow and maxTokens when the replacement only has sentinel
fallbacks. Custom models (via #mergeCustomModels) already had this
protection via ?? fallback; provider discoveries didn't.

Fixes the TUI showing 222222/8888 instead of the real context/token
limits for discovered models.

packages/coding-agent/CHANGELOG.md

@@ -19,6 +19,10 @@
 - Changed search truncation metadata/renderer output from match/result-based limits to file-based limits (`fileLimitReached`, `perFileLimitReached`) and updated truncation labels accordingly
 - Lowered `read.defaultLimit` default from `500` to `300` lines, and split the per-range context padding into asymmetric `RANGE_LEADING_CONTEXT_LINES = 1` / `RANGE_TRAILING_CONTEXT_LINES = 3` (was symmetric `RANGE_CONTEXT_LINES = 3`). Replay analysis over post-summarizer sessions (`scripts/session-stats/optimize_read_config.py`) showed that bare-path reads are over-provisioned at the median (file p50 = 220 lines) and that most follow-up reads are disjoint hops rather than adjacent extensions — so a smaller default plus narrower leading context reclaims tokens without measurably changing first-cover rate. Trailing context stays at 3 lines to keep anchor-stale recovery on narrow reads. Explicit `read.defaultLimit` overrides in settings are honoured unchanged.
 
+	
+### Fixed
+
+- Fixed model contextWindow and maxTokens defaulting to `UNK_CONTEXT_WINDOW` (222222) / `UNK_MAX_TOKENS` (8888) when cached or freshly-discovered provider models replace bundled models through `ModelRegistry.#mergeResolvedModels`. The merge now preserves the bundled model's values when the replacement only has sentinel fallbacks.
 ## [15.0.0] - 2026-05-13
 ### Breaking Changes
 

packages/coding-agent/src/config/model-registry.ts

@@ -18,6 +18,8 @@ import {
 	registerCustomApi,
 	type SimpleStreamOptions,
 	type ThinkingConfig,
+	UNK_CONTEXT_WINDOW,
+	UNK_MAX_TOKENS,
 	unregisterCustomApis,
 } from "@oh-my-pi/pi-ai";
 
@@ -1053,7 +1055,16 @@ export class ModelRegistry {
 			const key = `${replacementModel.provider}\u0000${replacementModel.id}`;
 			const existingIndex = indexByKey.get(key);
 			if (existingIndex !== undefined) {
-				merged[existingIndex] = replacementModel;
+				const existing = merged[existingIndex];
+				merged[existingIndex] = {
+					...replacementModel,
+					contextWindow:
+						replacementModel.contextWindow === UNK_CONTEXT_WINDOW
+							? existing.contextWindow
+							: replacementModel.contextWindow,
+					maxTokens:
+						replacementModel.maxTokens === UNK_MAX_TOKENS ? existing.maxTokens : replacementModel.maxTokens,
+				};
 			} else {
 				merged.push(replacementModel);
 				indexByKey.set(key, merged.length - 1);

packages/coding-agent/test/model-registry.test.ts

@@ -2121,4 +2121,48 @@ describe("ModelRegistry", () => {
 			expect(model?.isOAuth).toBeUndefined();
 		});
 	});
+
+	test("cached discovery with UNK contextWindow preserves bundled value", () => {
+		// Configure openai as a discoverable provider through models.json
+		writeRawModelsJson({
+			openai: {
+				baseUrl: "https://my-proxy.example.com/v1",
+				apiKey: "TEST_KEY",
+				api: "openai-completions",
+				discovery: { type: "openai-models-list" },
+				models: [],
+			},
+		});
+		// Pre-populate the cache with a model that has UNK sentinel values
+		// (simulating a discovery that didn't return limit.context)
+		writeModelCache<"openai-completions">(
+			"openai",
+			Date.now(),
+			[
+				{
+					id: "gpt-4o",
+					name: "GPT-4o",
+					api: "openai-completions",
+					provider: "openai",
+					baseUrl: "https://my-proxy.example.com/v1",
+					reasoning: false,
+					input: ["text"],
+					cost: { input: 0, output: 0, cacheRead: 0, cacheWrite: 0 },
+					contextWindow: 222_222, // UNK_CONTEXT_WINDOW
+					maxTokens: 8_888, // UNK_MAX_TOKENS
+				},
+			],
+			true,
+			cacheDbPath,
+		);
+		const registry = new ModelRegistry(authStorage, modelsJsonPath);
+		const model = registry.find("openai", "gpt-4o");
+
+		expect(model).toBeDefined();
+		// The bundled gpt-4o has a correct contextWindow, not the UNK sentinel
+		expect(model!.contextWindow).not.toBe(222_222);
+		expect(model!.contextWindow).toBeGreaterThan(100_000);
+		expect(model!.maxTokens).not.toBe(8_888);
+		expect(model!.maxTokens).toBeGreaterThan(1000);
+	});
 });