@gtrabanco/pi-nan-provider 1.0.1 → 1.2.1

package/src/fetch-models.ts

@@ -85,6 +85,41 @@ export const UNKNOWN_MODEL_LIMITS = { contextWindow: 128_000, maxTokens: 4_096 }
 /** Timeout for the live /models fetch; matches the pi-synthetic-provider precedent (~3s). */
 export const DEFAULT_MODELS_TIMEOUT_MS = 3_000;
 
+/**
+ * Capability data for models the docs serve but the static baseline must NOT
+ * carry (tier-gated), so the live `/models` path registers them with real
+ * documented values instead of the blind `UNKNOWN_MODEL_LIMITS` placeholder.
+ *
+ * `glm5.3` is premium-tier: a non-premium key must never see it in the static
+ * baseline (it cannot call it), but a premium key that discovers it live must
+ * get its real capabilities. Every value traces to
+ * https://nan.builders/docs/models (checked 2026-10-14; docs are the source of
+ * truth — models.dev lags):
+ * - Context 1M tokens, multimodal input text·image·audio·video (pi-representable
+ *   subset: text, image; pi's Model.input has no audio/video), output text,
+ *   tool calling, reasoning_effort low·medium·high·max.
+ * - Max output 131,072: models.dev documents that value for glm5.3; the docs
+ *   give the same 131K figure for the sibling glm5.3-flash, and do not state a
+ *   lower bound for glm5.3. Kept as the confirmed maximum rather than invented.
+ * - thinkingLevelMap { off: "low" }: the docs list low as glm5.3's shallowest
+ *   level, so `thinking: off` sends the shallowest supported effort. glm5.3 was
+ *   never measured (issue #16 covered glm5.3-flash, whose shallowest effective
+ *   value is "minimal"); "low" is the documented floor.
+ */
+export const LIVE_ONLY_MODEL_DATA: Readonly<Record<string, Omit<GeneratedModelEntry, "compat">>> = {
+	"glm5.3": {
+		id: "glm5.3",
+		name: "GLM 5.3",
+		reasoning: true,
+		input: ["text", "image"],
+		cost: { input: 0, output: 0, cacheRead: 0, cacheWrite: 0 },
+		contextWindow: 1_048_576,
+		maxTokens: 131_072,
+		reasoningEffortValues: ["low", "medium", "high", "max"],
+		thinkingLevelMap: { off: "low" },
+	},
+};
+
 /**
  * Live endpoint IDs classified as non-chat and excluded from the registered
  * model set. Keys are model ids; values are provenance reasons citing the
@@ -218,6 +253,10 @@ export function mergeLiveWithGenerated(
 	generated: readonly GeneratedModelEntry[] = NAN_GENERATED_MODELS,
 ): MergedCatalog {
 	const byId = new Map(generated.map((entry) => [entry.id, entry]));
+	// Every generated entry shares the same NaN/LiteLLM compat block; borrow it
+	// for the live-only documented entries so the gateway-behavior flags never
+	// drift between the static catalog and the live overlay.
+	const catalogCompat = generated.find((entry) => entry.compat)?.compat;
 	const models: Model<"openai-completions">[] = [];
 	const matched: string[] = [];
 	const unknown: string[] = [];
@@ -235,6 +274,16 @@ export function mergeLiveWithGenerated(
 			models.push(toModel(entry, source));
 			matched.push(id);
 		} else {
+			// Documented data for tier-gated live-only ids (glm5.3): the static
+			// baseline excludes them so a non-premium key never sees a model it
+			// cannot call, but a key that discovers one live gets its real
+			// capabilities instead of the blind placeholder below.
+			const liveOnlyEntry = LIVE_ONLY_MODEL_DATA[id];
+			if (liveOnlyEntry) {
+				models.push(toModel({ ...liveOnlyEntry, compat: catalogCompat }, source));
+				matched.push(id);
+				continue;
+			}
 			// Placeholder for uncatalogued live ids (e.g. new or undocumented):
 			// limits are the documented safe envelope and capabilities stay
 			// "unknown". The compat keys are the same gateway-behavior flags
@@ -443,6 +492,15 @@ export function buildOverlayMap(
 	for (const entry of generated) {
 		overlay.set(entry.id, { chat: toModel(entry, source) });
 	}
+	// Tier-gated live-only chat models with documented capabilities (glm5.3):
+	// absent from the static baseline so a non-premium key never sees them, but
+	// present here so a stored entry for one is corrected to the documented
+	// values instead of passing through stale. The store still decides which ids
+	// exist (issue #19); this map only decides their capabilities.
+	const catalogCompat = generated.find((entry) => entry.compat)?.compat;
+	for (const entry of Object.values(LIVE_ONLY_MODEL_DATA)) {
+		overlay.set(entry.id, { chat: toModel({ ...entry, compat: catalogCompat }, source) });
+	}
 	// Image models from the baseline (these live under a different api id
 	// so they cannot coexist as chat entries — ids are unique across types).
 	if (imageModels) {

package/src/index.ts

@@ -15,17 +15,13 @@
  *     0.84 alike), and the openai-completions streaming implementation is
  *     resolved dynamically — see provider-factory.ts for why.
  *
- *  2. MCP servers via pi's native registerMcpServer (pi >=0.99):
- *     - `nan-search` — the official NaN remote MCP server
- *       (https://api.nan.builders/mcp), exposing `mcp__nan_search__web_search`
- *       (pi >=0.99.2; 0.99.0-0.99.1 kept the hyphen: `mcp__nan-search__web_search`).
- *       Default: enabled. NAN_MCP_TOOLS=0 to disable.
+ *  2. MCP server via pi's native registerMcpServer (pi >=0.99):
  *     - `nan-media` — the community stdio media server
  *       (flux-2-klein / kokoro / whisper), exposing
  *       `mcp__nan_media__generate_image` etc. Default: enabled,
  *       NAN_MEDIA_MCP=0 to disable.
  *
- *     Both are session-scoped (visible in /mcp with source "extension").
+ *     It is session-scoped (visible in /mcp with source "extension").
  *     Exposure: `direct` — tools are declared to the model like built-ins.
  */
 
@@ -40,7 +36,7 @@ import {
 import { baselineModels } from "./fetch-models.ts";
 import { createNanCompatibleProvider, type OpenAICompatibleProviderConfig } from "./provider-factory.ts";
 import { PROVIDERS } from "./providers.ts";
-import { NAN_API_KEY_ENV, mcpToolsDisabled, mcpToolsEnvExplicit, resolveNanApiKey, tryResolveNanApiKeyViaRegistry, resolveStoredNanApiKey, NAN_MCP_TOOLS_ENV, webSearchBridgeEnabled } from "./mcp/api-key.ts";
+import { NAN_API_KEY_ENV, resolveNanApiKey, tryResolveNanApiKeyViaRegistry, resolveStoredNanApiKey } from "./mcp/api-key.ts";
 import { mediaMcpCommand, mediaMcpEnabled, mediaMcpEnvExplicit, mediaMcpEnvTruthy, mediaMcpTimeoutSec } from "./mcp/media-server.ts";
 import { registerMcpHostGuard } from "./mcp/host-support.ts";
 
@@ -95,26 +91,6 @@ async function registerProviderCompat(
 const NO_NATIVE_MCP_GUARD_MSG =
 	"pi-nan-provider 0.10+ requires pi >= 0.99 for MCP tools (native MCP); upgrade pi or stay on package 0.9.x";
 
-/** Register the official NaN web-search MCP server (session-scoped). */
-async function registerWebSearchMcpServer(pi: ExtensionAPI, apiKey: string | undefined): Promise<boolean> {
-	if (!webSearchBridgeEnabled()) return false;
-	if (!apiKey) {
-		console.warn(
-			"[pi-nan-provider] NAN_API_KEY is not set — web-search server not registered. " +
-				"Run `/login nan` in pi or export `NAN_API_KEY=<your-key>`.",
-		);
-		return false;
-	}
-	pi.registerMcpServer("nan-search", {
-		type: "http",
-		url: "https://api.nan.builders/mcp",
-		headers: { Authorization: `Bearer ${apiKey}` },
-		exposure: "direct",
-		description: "Web search through NaN's gateway (official NaN MCP server)",
-	});
-	return true;
-}
-
 /** Register the community media stdio MCP server (session-scoped). */
 function registerMediaMcpServer(pi: ExtensionAPI, apiKey: string | undefined): boolean {
 	if (!mediaMcpEnabled()) return false;
@@ -139,11 +115,11 @@ function registerMediaMcpServer(pi: ExtensionAPI, apiKey: string | undefined): b
 }
 
 /**
- * Register MCP servers via pi's native registerMcpServer.
+ * Register the MCP server via pi's native registerMcpServer.
  *
- * Session-scoped: servers appear in /mcp with source "extension", are
- * visible to the model (exposure: "direct"), and user mcp.json entries
- * with the same name take precedence.
+ * Session-scoped: the server appears in /mcp with source "extension", is
+ * visible to the model (exposure: "direct"), and a user mcp.json entry
+ * with the same name takes precedence.
  *
  * On pi <0.99 (registerMcpServer absent) this does NOT bridge — it
  * notifies once and skips (same loud-skip style as existing guards).
@@ -162,7 +138,6 @@ async function registerMcpServersNative(pi: ExtensionAPI): Promise<void> {
 	const storedKey = resolveStoredNanApiKey();
 	const apiKey = storedKey ?? process.env[NAN_API_KEY_ENV];
 
-	await registerWebSearchMcpServer(pi, apiKey);
 	registerMediaMcpServer(pi, apiKey);
 }
 

package/src/mcp/api-key.ts

@@ -1,8 +1,7 @@
-import { bridgeSource, resolveBridgeEnabled } from "./state.ts";
 import { readStoredCredential } from "@earendil-works/pi-coding-agent";
 
 /**
- * NaN API key resolution — shared across every surface (web search, media, usage).
+ * NaN API key resolution — shared across every surface (media, usage).
  *
  * Strategy: pi's stored credential (modelRegistry) wins → env var fallback.
  * Never logs or embeds the key beyond the Authorization header.
@@ -40,8 +39,6 @@ export function resolveStoredNanApiKey(): string | undefined {
 	}
 }
 
-/** Env var that overrides the official web_search bridge. */
-export const NAN_MCP_TOOLS_ENV = "NAN_MCP_TOOLS";
 /** Env var that overrides the media bridge. */
 export const NAN_MEDIA_MCP_ENV = "NAN_MEDIA_MCP";
 /** The NaN API key env var used everywhere. */
@@ -74,18 +71,6 @@ export async function tryResolveNanApiKeyViaRegistry(
 	}
 }
 
-/** Whether MCP tool registration is disabled via env (NAN_MCP_TOOLS=0|false|off). */
-export function mcpToolsDisabled(): boolean {
-	const value = process.env[NAN_MCP_TOOLS_ENV]?.trim().toLowerCase();
-	return value === "0" || value === "false" || value === "off";
-}
-
-/** Whether NAN_MCP_TOOLS is explicitly set (any value). */
-export function mcpToolsEnvExplicit(): boolean {
-	const value = process.env[NAN_MCP_TOOLS_ENV];
-	return value !== undefined && value.trim() !== "";
-}
-
 /** Whether NAN_MEDIA_MCP is explicitly set (any value). */
 export function mediaMcpEnvExplicit(): boolean {
 	const value = process.env[NAN_MEDIA_MCP_ENV];
@@ -97,13 +82,3 @@ export function mediaMcpEnvTruthy(): boolean {
 	const value = process.env[NAN_MEDIA_MCP_ENV]?.trim().toLowerCase();
 	return value === "1" || value === "true" || value === "on";
 }
-
-/** Source of the effective web-search enablement (env / persisted / default). */
-export function webSearchBridgeSource(): import("./state.ts").BridgeSource {
-	return bridgeSource("webSearch", mcpToolsEnvExplicit());
-}
-
-/** Whether the web-search bridge is enabled by gate (env → persisted → default). */
-export function webSearchBridgeEnabled(): boolean {
-	return resolveBridgeEnabled("webSearch", mcpToolsEnvExplicit(), !mcpToolsDisabled(), true);
-}

package/src/mcp/media-server.ts

@@ -27,7 +27,7 @@ export function mediaMcpTimeoutSec(): number {
 
 export const NAN_MEDIA_MCP_ENV = "NAN_MEDIA_MCP";
 
-export const DEFAULT_NAN_MEDIA_MCP_VERSION = "1.1.2";
+export const DEFAULT_NAN_MEDIA_MCP_VERSION = "1.3.0";
 export const DEFAULT_MEDIA_MCP_TIMEOUT_MS = 120_000;
 export const NAN_MEDIA_MCP_VERSION_ENV = "NAN_MEDIA_MCP_VERSION";
 export const NAN_MEDIA_MCP_COMMAND_ENV = "NAN_MEDIA_MCP_COMMAND";

package/src/mcp/state.ts

@@ -17,8 +17,6 @@ import { getAgentDir } from "@earendil-works/pi-coding-agent";
 export const NAN_STATE_FILE = "nan-provider.json";
 
 export interface NanProviderState {
-	/** Persisted enablement of the official web_search MCP bridge. */
-	webSearch?: boolean;
 	/** Persisted enablement of the community nan-mcp-server media bridge. */
 	mediaMcp?: boolean;
 }
@@ -27,7 +25,6 @@ export type BridgeKey = keyof NanProviderState;
 
 /** Server name each bridge registers with `pi.registerMcpServer()`. Single source of truth. */
 export const MCP_SERVER_NAMES: Record<BridgeKey, string> = {
-	webSearch: "nan-search",
 	mediaMcp: "nan-media",
 };
 

package/src/reasoning-only-stream-guard.ts

@@ -2,8 +2,13 @@
  * NaN reasoning-only stream guard (issue #21).
  *
  * Problem: NaN's gateway closes a streaming turn that produced ONLY reasoning
- * (no content, no tool calls) at 60,000 reasoning chars or 420 seconds:
- * `finish_reason: "length"` on an empty delta, followed by a final usage chunk
+ * (no content, no tool calls) once it has produced more than 45,000 reasoning
+ * characters, or 420 seconds without visible output (600 seconds once the turn
+ * has demonstrably exceeded that character count). The reasoning-character
+ * budget also scales with prompt size: ~150,000 characters of headroom plus
+ * half the prompt's size, capped at 300,000
+ * (https://nan.builders/docs/models#reasoning-only-stream-limit,
+ * checked 2026-10-14). The cut is:
  * (estimated, unbilled) carrying `nan_truncation: { reason: "reasoning_only_stream" }`.
  * Through pi-ai this maps to `stopReason: "length"` with no text — indistinguishable
  * from a real output-limit cut, never retried, and the turn is silently wasted.
@@ -72,7 +77,7 @@ export function isReasoningOnlyStreamChunk(chunk: unknown): boolean {
 const REASONING_ONLY_REWRITE: Partial<AssistantMessage> = {
 	stopReason: "error" as const,
 	errorMessage:
-		"NaN timed out the stream before any answer content (reasoning-only stream guard: 420 s / 60,000-character limit; the turn was not billed). Original finish_reason: length.",
+		"NaN timed out the stream before any answer content (reasoning-only stream guard: the 420 s / 45,000-reasoning-character limit; the turn was not billed). Original finish_reason: length.",
 };
 
 /**