copilot-relay 0.3.7 → 0.3.8

package/README.md

@@ -117,16 +117,19 @@ for live model discovery, compatible thinking efforts, and copy-paste updates.
 | Level | Logs |
 | --- | --- |
 | `error` | Startup, preflight, and request failures |
-| `info` | Errors plus startup status, preflight status, request IDs, upstream lifecycle, and local HTTP status codes |
-| `debug` | Info plus model routing summaries, Copilot upstream timings, and request payloads |
+| `info` | Errors plus startup/preflight status, request IDs, model/effort summaries, upstream lifecycle, and local HTTP status codes |
+| `debug` | Info plus detailed Copilot timings and request payloads |
 
 Any other `logLevel` value is invalid and stops startup.
 
-Valid `thinkEffort`: `none`, `low`, `medium`, `high`, `xhigh`, `max`.
+Valid `thinkEffort` defaults: `low`, `medium`, `high`, `xhigh`, `max`.
 It is a fallback, not a forced override: Claude Code's `output_config.effort`
 wins, followed by the legacy `reasoning_effort` field, then `thinkEffort` when
 neither request field supplies a value. The selected effort is preserved through
 all model and WebSearch passes; existing config values are not rewritten.
+`none` and other invalid explicit defaults fail startup before authentication or
+config write-back, with a clear list of valid choices. They are not silently
+replaced with `max`. Explicit request-level `none` remains model-dependent.
 
 `upstreamTimeoutSeconds` controls the maximum time a single Claude request can
 spend waiting on upstream Copilot calls, including chat, Responses, preflight,
@@ -212,7 +215,10 @@ health probe, so a relay that cannot answer `/healthz` never reports success.
 
 ## Logging
 
-At `debug`, every model request logs the requested model, upstream model, requested think effort, requested thinking, and effective think effort.
+At `info`, every model request logs the requested model, upstream model, requested
+think effort, requested thinking, and effective think effort. Missing request
+effort is shown as `unset`, distinct from explicit `none`; full normal request
+payloads remain at `debug`.
 
 Upstream failures are logged at `error` with full request and response context in the same log file.
 

package/config.default.yaml

@@ -26,10 +26,9 @@ claudeSetup: true
 
 # Log verbosity:
 #   error - startup/preflight/request failures only
-#   info  - error logs plus startup status, preflight status, and local HTTP
-#           status codes
-#   debug - info logs plus model routing summaries, Copilot upstream timings,
-#           and request payloads
+#   info  - error logs plus startup/preflight status, model/effort summaries,
+#           and local HTTP status codes
+#   debug - info logs plus detailed Copilot timings and request payloads
 # Any other value is invalid and stops startup.
 logLevel: info
 
@@ -41,7 +40,7 @@ logRetentionDays: 3
 
 # Fallback upstream effort when Claude Code does not specify one.
 # Request output_config.effort wins, followed by legacy reasoning_effort.
-# Valid values: none, low, medium, high, xhigh, max.
+# Valid values: low, medium, high, xhigh, max.
 thinkEffort: max
 
 # Max seconds to wait for one Claude request's upstream Copilot calls.

package/dist/main.js

@@ -541,12 +541,20 @@ const runtimeState = {};
 
 //#endregion
 //#region src/lib/models.ts
+const configurableReasoningEfforts = [
+	"low",
+	"medium",
+	"high",
+	"xhigh",
+	"max"
+];
 const defaultReasoningEffort = "max";
 const defaultModelRouting = {
 	gptModel: "gpt-6-astra",
 	opusModel: "claude-opus-5"
 };
-const isReasoningEffort = (value) => value === "none" || value === "low" || value === "medium" || value === "high" || value === "xhigh" || value === "max";
+const isConfiguredReasoningEffort = (value) => configurableReasoningEfforts.some((effort) => effort === value);
+const isReasoningEffort = (value) => value === "none" || isConfiguredReasoningEffort(value);
 const invalidRequestEffort = (message) => new HTTPError(message, new Response(JSON.stringify({
 	type: "error",
 	error: {
@@ -736,10 +744,19 @@ const normalizeCopilotBaseUrl = (value) => {
 	if (parsed.username || parsed.password) throw new Error(copilotBaseUrlCredentialsMessage);
 	return trimmed;
 };
+var InvalidThinkEffortError = class extends Error {
+	constructor() {
+		super(`Invalid thinkEffort. Valid values: ${configurableReasoningEfforts.join(", ")}. "none" is not allowed as a configured default.`);
+		this.name = "InvalidThinkEffortError";
+	}
+};
 const normalizeThinkEffort = (value) => {
-	if (typeof value !== "string") return;
-	const normalized = value.toLowerCase() === "minimal" ? "low" : value.toLowerCase();
-	return isReasoningEffort(normalized) ? normalized : void 0;
+	if (value === void 0) return void 0;
+	if (typeof value === "string") {
+		const normalized = value.toLowerCase() === "minimal" ? "low" : value.toLowerCase();
+		if (isConfiguredReasoningEffort(normalized)) return normalized;
+	}
+	throw new InvalidThinkEffortError();
 };
 const normalizeUpstreamTimeoutSeconds = (value) => {
 	if (value === void 0) return;
@@ -858,16 +875,15 @@ const serializeConfig = (config) => [
 	"",
 	"# Log verbosity:",
 	"#   error - startup/preflight/request failures only",
-	"#   info  - error logs plus startup status, preflight status, and local HTTP",
-	"#           status codes",
-	"#   debug - info logs plus model routing summaries, Copilot upstream timings,",
-	"#           and request payloads",
+	"#   info  - error logs plus startup/preflight status, model/effort summaries,",
+	"#           and local HTTP status codes",
+	"#   debug - info logs plus detailed Copilot timings and request payloads",
 	`logLevel: ${config.logLevel}`,
 	"",
 	"# Number of days to keep files in ~/.copilot-relay/logs.",
 	`logRetentionDays: ${config.logRetentionDays}`,
 	"",
-	"# Fallback effort when the request omits it: none, low, medium, high, xhigh, max.",
+	`# Fallback effort when the request omits it: ${configurableReasoningEfforts.join(", ")}.`,
 	`thinkEffort: ${config.thinkEffort}`,
 	"",
 	"# Max seconds for one request's upstream calls; 0 disables the relay deadline.",
@@ -926,7 +942,10 @@ const watchAppConfig = (onReload) => {
 			if (nextMtime === lastMtime) return;
 			lastMtime = nextMtime;
 			onReload(await readAppConfig());
-		} catch {}
+		} catch (error) {
+			if (error instanceof InvalidThinkEffortError) log.error(`${error.message} Keeping the previous runtime settings.`);
+			else log.error(`Could not reload config. Check ${paths.configPath} for invalid values or file errors.`);
+		}
 	}, 1e3);
 	if (typeof timer.unref === "function") timer.unref();
 	getConfigMtime().then((mtime) => {
@@ -2081,7 +2100,7 @@ const messagesIncludeImage = (messages) => messages.some((msg) => typeof msg.con
 const createChatCompletions = async (config, payload, options = {}) => {
 	const client = options.client ?? "generic";
 	const requestedModel = options.requestedModel ?? payload.model;
-	const requestedThinkEffort = options.requestedThinkEffort ?? getRequestReasoningEffort(payload) ?? "none";
+	const requestedThinkEffort = options.requestedThinkEffort ?? getRequestReasoningEffort(payload) ?? "unset";
 	const reasoningEffort = getRequestedReasoningEffort(payload);
 	const requestedThinking = options.requestedThinking ?? "none";
 	const signal = createCopilotRequestSignal(options.signal, options.timeoutMs);
@@ -2103,7 +2122,7 @@ const createChatCompletions = async (config, payload, options = {}) => {
 	const enableVision = messagesIncludeImage(compatiblePayload.messages);
 	const initiator = isAgentInitiator(compatiblePayload.messages);
 	const requestPayload = buildRequestPayload(compatiblePayload);
-	log.debug([
+	log.info(sanitizeTerminalString([
 		"Model request",
 		`client=${client}`,
 		`requested_model=${requestedModel}`,
@@ -2111,8 +2130,8 @@ const createChatCompletions = async (config, payload, options = {}) => {
 		`upstream_model=${compatiblePayload.model}`,
 		`requested_think_effort=${requestedThinkEffort}`,
 		`requested_thinking=${requestedThinking}`,
-		`effective_think_effort=${requestPayload.reasoning_effort ?? "none"}`
-	].join(" "));
+		`effective_think_effort=${requestPayload.reasoning_effort ?? "unset"}`
+	].join(" ")));
 	log.debug("Full request payload", { payload: requestPayload });
 	if (useResponsesApi) return completeResponse(await createResponses(provider, compatiblePayload, {
 		vision: enableVision,
@@ -3216,7 +3235,7 @@ const createTokenCountModel = (modelId, tokenizer = "o200k_base") => ({
 	id: modelId
 });
 const isNonStreamingResponse = (response) => typeof response === "object" && response !== null && Object.hasOwn(response, "choices");
-const getClaudeRequestedThinkEffort = (payload) => getRequestReasoningEffort(payload) ?? "none";
+const getClaudeRequestedThinkEffort = (payload) => getRequestReasoningEffort(payload) ?? "unset";
 const getClaudeRequestedThinking = (payload) => {
 	if (!payload.thinking) return "none";
 	return [`type:${payload.thinking.type}`, `budget:${payload.thinking.budget_tokens ?? "none"}`].join(",");
@@ -3645,6 +3664,10 @@ const startServer = (config) => new Promise((resolve$1, reject) => {
 const canCloseConnections = (server) => typeof server.closeIdleConnections === "function" && typeof server.closeAllConnections === "function";
 async function startRelay(appConfig) {
 	appConfig ??= await readAppConfig();
+	appConfig = {
+		...appConfig,
+		thinkEffort: normalizeThinkEffort(appConfig.thinkEffort) ?? defaultReasoningEffort
+	};
 	setLogLevel(appConfig.logLevel);
 	await cleanupLogs(appConfig.logRetentionDays);
 	const claudeConfigPath$1 = claudeConfigPath;