@dianshuv/copilot-api 0.20.7 → 0.21.1

package/README.md

@@ -15,7 +15,7 @@
 - **Graceful shutdown**: 4-phase shutdown sequence — stops accepting requests, waits for in-flight requests to complete, sends abort signal, then force-closes. Configurable via `--shutdown-graceful-wait` and `--shutdown-abort-wait`.
 - **Stream repetition detection**: Detects when models get stuck in repetitive output loops using KMP-based pattern matching and logs a warning.
 - **Stale request reaping**: Automatically force-fails requests that exceed a configurable maximum age (default 600s) to prevent resource leaks.
-- **PostHog analytics**: Optional PostHog Cloud integration (`--posthog-key`) sends per-request token usage events for long-term trend analysis. Free tier (1M events/month) is more than sufficient for individual use.
+- **PostHog analytics**: Optional PostHog Cloud integration (`--posthog-key`) sends per-request token usage and estimated GitHub AI credit consumption for long-term trend analysis. Credit estimates use the live per-model prices returned by Copilot's `/models` endpoint, including default/long-context and cache pricing; no price table is hardcoded. Free tier (1M events/month) is more than sufficient for individual use.
 - **GitHub Copilot CLI emulation**: All upstream requests to GitHub — device-flow `login`, `/copilot_internal/user` bootstrap, and CAPI model/chat calls — carry the official GitHub Copilot CLI's (`@github/copilot`) identity: its `copilot-integration-id` (`copilot-developer-cli`), `editor-version`/`user-agent` (`copilot/<version>`), `x-github-api-version`, and a persistent `x-client-machine-id` (stored at `~/.local/share/copilot-api/machine_id`). The CAPI Bearer is the GitHub OAuth token itself — the CLI does not perform a separate token exchange. The `login` flow uses the CLI's own OAuth app, so **new** logins request the `read:user`, `read:org`, `repo`, and `gist` scopes; existing tokens keep working without re-authentication.
 
 ## Quick Start
@@ -67,7 +67,7 @@ make down
 
 ### Hidden Models
 
-By default the proxy hides 38 stale / duplicate / unused upstream model ids from
+By default the proxy hides 40 stale / duplicate / unused upstream model ids from
 its **listing surfaces** (`/v1/models` and the startup banner).
 
 **Important**: this is a *display* filter only. The remaining POST endpoints
@@ -87,9 +87,9 @@ Currently hidden (grouped):
 - **GPT-4.1 family** — `gpt-4.1`, `gpt-4.1-2025-04-14`, `gpt-41-copilot`
 - **Older / smaller GPT-5** — `gpt-5-mini`, `gpt-5.3-codex`, `gpt-5.4`, `gpt-5.4-mini`, `gpt-5.4-nano`, `gpt-5.5`
 - **All embeddings** — `text-embedding-ada-002`, `text-embedding-3-small`, `text-embedding-3-small-inference`
-- **Older Gemini** — `gemini-2.5-pro`, `gemini-3-flash-preview`, `gemini-3.5-flash`
+- **Older Gemini** — `gemini-2.5-pro`, `gemini-3-flash-preview`, `gemini-3.1-pro-preview`, `gemini-3.5-flash`
 - **Older / variant Claude** — `claude-opus-4.5`, `claude-opus-4.6`, `claude-opus-4.7`, `claude-opus-4.8`, `claude-opus-4.7-high`, `claude-opus-4.7-xhigh`, `claude-sonnet-4.5`, `claude-sonnet-4.6`
-- **Special-purpose** — `mai-code-1-flash-internal`, `mai-code-1-flash-picker`, `trajectory-compaction`
+- **Special-purpose** — `mai-code-1-flash-internal`, `mai-code-1-flash-picker`, `mai-code-1.1-flash`, `trajectory-compaction`
 
 ## API Endpoints
 
@@ -120,6 +120,14 @@ Currently hidden (grouped):
 | `/history` | GET | Request history Web UI with token analytics (enabled by default) |
 | `/history/api/*` | GET/DELETE | History API endpoints |
 
+### PostHog AI credit analytics
+
+When `--posthog-key` is configured, successful `copilot_api_request` events with upstream usage data include an `ai_credits` property. The proxy calculates it from the request's fresh input, output, cache-read, and cache-write token counts using the selected model's `billing.token_prices` metadata cached from Copilot's `/models` response. If total prompt tokens exceed the model's default threshold, the calculation uses its `long_context` prices.
+
+To chart daily consumption in PostHog, filter to `copilot_api_request`, aggregate **sum of `ai_credits`**, set the interval to **day**, and optionally break down by `model`. Events omit `ai_credits` when Copilot provides no usage data or usable billing metadata; zero-priced models report `0`.
+
+`ai_credits` is a per-request estimate from token categories exposed by the upstream response. If an upstream API bills a token category it does not report (for example, an undisclosed cache write), the GitHub billing total can be higher than the estimate.
+
 ## Auto-Truncate
 
 When enabled (default), auto-truncate automatically compacts conversation history when it exceeds the model's token limit. This prevents request failures due to context overflow.

package/dist/main.mjs

@@ -1011,7 +1011,7 @@ const logout = defineCommand({
 
 //#endregion
 //#region package.json
-var version = "0.20.7";
+var version = "0.21.1";
 
 //#endregion
 //#region src/lib/event-loop-lag.ts
@@ -1596,6 +1596,8 @@ function captureRequest(params) {
 	if (params.cachedInputTokens !== void 0) properties.cached_input_tokens = params.cachedInputTokens;
 	if (params.cacheCreationInputTokens !== void 0) properties.cache_creation_input_tokens = params.cacheCreationInputTokens;
 	if (params.totalInputTokens !== void 0) properties.total_input_tokens = params.totalInputTokens;
+	const aiCredits = calculateAiCredits(params);
+	if (aiCredits !== void 0) properties.ai_credits = aiCredits;
 	if (params.stopReason !== void 0) properties.stop_reason = params.stopReason;
 	if (params.status !== void 0) properties.status = params.status;
 	if (params.copilotErrorCode !== void 0) properties.copilot_error_code = params.copilotErrorCode;
@@ -1613,6 +1615,47 @@ function captureRequest(params) {
 		properties
 	});
 }
+function calculateAiCredits(params) {
+	const tokenPrices = params.tokenPrices;
+	if (!isModelTokenPrices(tokenPrices)) return void 0;
+	const cachedInputTokens = params.cachedInputTokens ?? 0;
+	const cacheCreationInputTokens = params.cacheCreationInputTokens ?? 0;
+	const totalInputTokens = params.totalInputTokens ?? params.inputTokens + cachedInputTokens + cacheCreationInputTokens;
+	const defaultMaxPromptTokens = tokenPrices.default.max_prompt_tokens;
+	const prices = tokenPrices.long_context && defaultMaxPromptTokens !== void 0 && isNonNegativeFiniteNumber(defaultMaxPromptTokens) && totalInputTokens > defaultMaxPromptTokens ? tokenPrices.long_context : tokenPrices.default;
+	const tokens = [
+		params.inputTokens,
+		params.outputTokens,
+		cachedInputTokens,
+		cacheCreationInputTokens,
+		totalInputTokens
+	];
+	const rates = [
+		prices.input_price,
+		prices.output_price,
+		prices.cache_read_price,
+		prices.cache_write_price
+	];
+	if (!tokens.every((token) => isNonNegativeFiniteNumber(token)) || !rates.every((rate) => isNonNegativeFiniteNumber(rate)) || !isNonNegativeFiniteNumber(tokenPrices.batch_size)) return;
+	if (tokenPrices.batch_size === 0) return rates.every((rate) => rate === 0) ? 0 : void 0;
+	return (params.inputTokens * prices.input_price + params.outputTokens * prices.output_price + cachedInputTokens * prices.cache_read_price + cacheCreationInputTokens * prices.cache_write_price) / tokenPrices.batch_size;
+}
+function isModelTokenPrices(value) {
+	if (!isRecord$1(value)) return false;
+	if (!isNonNegativeFiniteNumber(value.batch_size)) return false;
+	if (!isModelTokenPriceTier(value.default)) return false;
+	return value.long_context === void 0 || isModelTokenPriceTier(value.long_context);
+}
+function isModelTokenPriceTier(value) {
+	if (!isRecord$1(value)) return false;
+	return isNonNegativeFiniteNumber(value.cache_read_price) && isNonNegativeFiniteNumber(value.cache_write_price) && isNonNegativeFiniteNumber(value.input_price) && isNonNegativeFiniteNumber(value.output_price) && (value.max_prompt_tokens === void 0 || isNonNegativeFiniteNumber(value.max_prompt_tokens));
+}
+function isRecord$1(value) {
+	return typeof value === "object" && value !== null && !Array.isArray(value);
+}
+function isNonNegativeFiniteNumber(value) {
+	return typeof value === "number" && Number.isFinite(value) && value >= 0;
+}
 async function shutdownPostHog() {
 	if (!client) return;
 	try {
@@ -2458,6 +2501,7 @@ const HIDDEN_MODEL_IDS = new Set([
 	"text-embedding-3-small-inference",
 	"gemini-2.5-pro",
 	"gemini-3-flash-preview",
+	"gemini-3.1-pro-preview",
 	"gemini-3.5-flash",
 	"claude-opus-4.5",
 	"claude-opus-4.6",
@@ -2469,6 +2513,7 @@ const HIDDEN_MODEL_IDS = new Set([
 	"claude-sonnet-4.6",
 	"mai-code-1-flash-internal",
 	"mai-code-1-flash-picker",
+	"mai-code-1.1-flash",
 	"trajectory-compaction"
 ]);
 function isHiddenModel(id, showAll) {
@@ -4062,14 +4107,6 @@ function recordStreamError(opts) {
 
 //#endregion
 //#region src/routes/tracker-mutations.ts
-/**
-* TUI tracker mutations and analytics completion.
-*
-* All in-place updates to the TUI request tracker (model, status, resolved
-* model) and the success/failure terminal transitions. `completeTracking`
-* additionally emits a PostHog analytics event for successful requests; error
-* paths emit their PostHog events through `observability-recording.ts`.
-*/
 /** Helper to update tracker model */
 function updateTrackerModel(trackingId, model, resolvedModel) {
 	if (!trackingId) return;
@@ -4119,6 +4156,7 @@ function completeTracking(trackingId, inputTokens, outputTokens, queueWaitMs, re
 		cachedInputTokens: cache?.cachedInputTokens,
 		cacheCreationInputTokens: cache?.cacheCreationInputTokens,
 		totalInputTokens: cache?.totalInputTokens,
+		tokenPrices: analytics.tokenPrices,
 		stopReason: analytics.stopReason
 	});
 }
@@ -4265,7 +4303,7 @@ async function executeRequest(opts) {
 			signal: abort.signal
 		}));
 		ctx.queueWaitMs = queueWaitMs;
-		if (isNonStreaming(response)) return handleNonStreamingResponse(c, response, ctx, payload);
+		if (isNonStreaming(response)) return handleNonStreamingResponse(c, response, ctx, payload, selectedModel);
 		consola.debug("Streaming response");
 		updateTrackerStatus(ctx.trackingId, "streaming");
 		return streamSSE(c, async (stream) => {
@@ -4274,6 +4312,7 @@ async function executeRequest(opts) {
 				stream,
 				response,
 				payload,
+				selectedModel,
 				ctx
 			});
 		});
@@ -4299,7 +4338,7 @@ async function logTokenCount(payload, selectedModel) {
 		consola.debug("Failed to calculate token count:", error);
 	}
 }
-function handleNonStreamingResponse(c, originalResponse, ctx, payload) {
+function handleNonStreamingResponse(c, originalResponse, ctx, payload, selectedModel) {
 	consola.debug("Non-streaming response:", JSON.stringify(originalResponse));
 	let response = originalResponse;
 	if (state.verbose && ctx.truncateResult?.wasCompacted && response.choices[0]?.message.content) {
@@ -4354,6 +4393,7 @@ function handleNonStreamingResponse(c, originalResponse, ctx, payload) {
 		cachedInputTokens,
 		cacheCreationInputTokens: 0,
 		totalInputTokens: usage?.prompt_tokens,
+		tokenPrices: usage ? selectedModel?.billing?.token_prices : void 0,
 		stopReason: choice.finish_reason
 	});
 	return c.json(echoResponseBody(response, ctx));
@@ -4386,6 +4426,7 @@ function createStreamAccumulator() {
 		outputTokens: 0,
 		cachedTokens: 0,
 		reasoningTokens: 0,
+		usageObserved: false,
 		finishReason: "",
 		content: "",
 		toolCalls: [],
@@ -4393,7 +4434,7 @@ function createStreamAccumulator() {
 	};
 }
 async function handleStreamingResponse(opts) {
-	const { stream, response, payload, ctx } = opts;
+	const { stream, response, payload, selectedModel, ctx } = opts;
 	const acc = createStreamAccumulator();
 	const checkRepetition = createStreamRepetitionChecker(`openai:${payload.model}`);
 	try {
@@ -4427,7 +4468,8 @@ async function handleStreamingResponse(opts) {
 			stream: true,
 			durationMs: Date.now() - ctx.startTime,
 			stopReason: acc.finishReason || void 0,
-			toolCount: payload.tools?.length ?? 0
+			toolCount: payload.tools?.length ?? 0,
+			tokenPrices: acc.usageObserved ? selectedModel?.billing?.token_prices : void 0
 		}, ctx.timings, {
 			cachedInputTokens: acc.cachedTokens,
 			cacheCreationInputTokens: 0,
@@ -4499,6 +4541,7 @@ async function accumulateAndEchoChunk(chunk, acc, checkRepetition, ctx, stream)
 function accumulateParsedChunk(parsed, acc, checkRepetition) {
 	if (parsed.model && !acc.model) acc.model = parsed.model;
 	if (parsed.usage) {
+		acc.usageObserved = true;
 		acc.inputTokens = parsed.usage.prompt_tokens;
 		acc.outputTokens = parsed.usage.completion_tokens;
 		acc.cachedTokens = parsed.usage.prompt_tokens_details?.cached_tokens ?? 0;
@@ -7899,11 +7942,12 @@ async function handleDirectAnthropicCompletion(c, anthropicPayload, ctx, initiat
 						stream,
 						response,
 						anthropicPayload: effectivePayload,
+						selectedModel,
 						ctx
 					});
 				});
 			}
-			return handleDirectAnthropicNonStreamingResponse(c, recoverLeakedToolCallsInResponse(response, toolNameSet(effectivePayload.tools)), ctx, truncateResult, effectivePayload);
+			return handleDirectAnthropicNonStreamingResponse(c, recoverLeakedToolCallsInResponse(response, toolNameSet(effectivePayload.tools)), ctx, truncateResult, effectivePayload, selectedModel);
 		}
 		consola.debug("[RateLimiter] Request queued past grace; opening keepalive stream (direct Anthropic)");
 		updateTrackerStatus(ctx.trackingId, "streaming");
@@ -7919,6 +7963,7 @@ async function handleDirectAnthropicCompletion(c, anthropicPayload, ctx, initiat
 						stream,
 						response,
 						anthropicPayload: effectivePayload,
+						selectedModel,
 						ctx
 					});
 				},
@@ -7973,7 +8018,7 @@ function logPayloadSizeInfoAnthropic(payload, model) {
 /**
 * Handle non-streaming direct Anthropic response
 */
-function handleDirectAnthropicNonStreamingResponse(c, response, ctx, truncateResult, payload) {
+function handleDirectAnthropicNonStreamingResponse(c, response, ctx, truncateResult, payload, selectedModel) {
 	consola.debug("Non-streaming response from Copilot (direct Anthropic):", JSON.stringify(response).slice(-400));
 	recordResponse(ctx.historyId, {
 		success: true,
@@ -8026,6 +8071,7 @@ function handleDirectAnthropicNonStreamingResponse(c, response, ctx, truncateRes
 		cachedInputTokens: cacheRead,
 		cacheCreationInputTokens: cacheCreation,
 		totalInputTokens,
+		tokenPrices: selectedModel?.billing?.token_prices,
 		stopReason: response.stop_reason ?? void 0
 	});
 	let finalResponse = response;
@@ -8071,7 +8117,7 @@ function echoForwardData(forwardData, eventType, ctx) {
 * Handle streaming direct Anthropic response (passthrough SSE events)
 */
 async function handleDirectAnthropicStreamingResponse(opts) {
-	const { stream, response, anthropicPayload, ctx } = opts;
+	const { stream, response, anthropicPayload, selectedModel, ctx } = opts;
 	const acc = createAnthropicStreamAccumulator();
 	const checkRepetition = createStreamRepetitionChecker(`anthropic:${anthropicPayload.model}`);
 	const serverToolFilter = createServerToolBlockFilter();
@@ -8110,7 +8156,8 @@ async function handleDirectAnthropicStreamingResponse(opts) {
 			stream: true,
 			durationMs: Date.now() - ctx.startTime,
 			stopReason: acc.stopReason || void 0,
-			toolCount: anthropicPayload.tools?.length ?? 0
+			toolCount: anthropicPayload.tools?.length ?? 0,
+			tokenPrices: selectedModel?.billing?.token_prices
 		}, ctx.timings, {
 			cachedInputTokens: acc.cacheReadInputTokens,
 			cacheCreationInputTokens: acc.cacheCreationInputTokens,
@@ -9041,7 +9088,8 @@ const handleResponses = async (c) => {
 							model: finalResult.model || model,
 							stream: true,
 							durationMs: Date.now() - startTime,
-							toolCount: tools.length
+							toolCount: tools.length,
+							tokenPrices: usage ? selectedModel?.billing?.token_prices : void 0
 						}, ctx.timings, {
 							cachedInputTokens,
 							cacheCreationInputTokens: 0,
@@ -9098,7 +9146,8 @@ const handleResponses = async (c) => {
 			model: result.model || model,
 			stream: false,
 			durationMs: Date.now() - startTime,
-			toolCount: tools.length
+			toolCount: tools.length,
+			tokenPrices: usage ? selectedModel?.billing?.token_prices : void 0
 		}, ctx.timings, {
 			cachedInputTokens,
 			cacheCreationInputTokens: 0,

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@dianshuv/copilot-api",
-  "version": "0.20.7",
+  "version": "0.21.1",
   "description": "Turn GitHub Copilot into OpenAI/Anthropic API compatible server. Usable with Claude Code!",
   "author": "dianshuv",
   "type": "module",