feature about responses api, also introduces a new configuration system, structured logging

README.md

@@ -185,11 +185,12 @@ The server exposes several endpoints to interact with the Copilot API. It provid
 
 These endpoints mimic the OpenAI API structure.
 
-| Endpoint                    | Method | Description                                               |
-| --------------------------- | ------ | --------------------------------------------------------- |
-| `POST /v1/chat/completions` | `POST` | Creates a model response for the given chat conversation. |
-| `GET /v1/models`            | `GET`  | Lists the currently available models.                     |
-| `POST /v1/embeddings`       | `POST` | Creates an embedding vector representing the input text.  |
+| Endpoint                    | Method | Description                                                      |
+| --------------------------- | ------ | ---------------------------------------------------------------- |
+| `POST /v1/responses`        | `POST` | Most advanced interface for generating model responses.          |
+| `POST /v1/chat/completions` | `POST` | Creates a model response for the given chat conversation.        |
+| `GET /v1/models`            | `GET`  | Lists the currently available models.                            |
+| `POST /v1/embeddings`       | `POST` | Creates an embedding vector representing the input text.         |
 
 ### Anthropic Compatible Endpoints
 

src/lib/config.ts

@@ -0,0 +1,97 @@
+import consola from "consola"
+import fs from "node:fs"
+
+import { PATHS } from "./paths"
+
+export interface AppConfig {
+  extraPrompts?: Record<string, string>
+  smallModel?: string
+  modelReasoningEfforts?: Record<string, "minimal" | "low" | "medium" | "high">
+}
+
+const defaultConfig: AppConfig = {
+  extraPrompts: {
+    "gpt-5-codex": `
+## Tool use
+- You have access to many tools. If a tool exists to perform a specific task, you MUST use that tool instead of running a terminal command to perform that task.
+### Bash tool
+When using the Bash tool, follow these rules:
+- always run_in_background set to false, unless you are running a long-running command (e.g., a server or a watch command).
+### BashOutput tool
+When using the BashOutput tool, follow these rules:
+- Only Bash Tool run_in_background set to true, Use BashOutput to read the output later
+### TodoWrite tool
+When using the TodoWrite tool, follow these rules:
+- Skip using the TodoWrite tool for tasks with three or fewer steps.
+- Do not make single-step todo lists.
+- When you made a todo, update it after having performed one of the sub-tasks that you shared on the todo list.
+## Special user requests
+- If the user makes a simple request (such as asking for the time) which you can fulfill by running a terminal command (such as 'date'), you should do so.
+`,
+  },
+  smallModel: "gpt-5-mini",
+  modelReasoningEfforts: {
+    "gpt-5-mini": "low",
+  },
+}
+
+let cachedConfig: AppConfig | null = null
+
+function ensureConfigFile(): void {
+  try {
+    fs.accessSync(PATHS.CONFIG_PATH, fs.constants.R_OK | fs.constants.W_OK)
+  } catch {
+    fs.mkdirSync(PATHS.APP_DIR, { recursive: true })
+    fs.writeFileSync(
+      PATHS.CONFIG_PATH,
+      `${JSON.stringify(defaultConfig, null, 2)}\n`,
+      "utf8",
+    )
+    try {
+      fs.chmodSync(PATHS.CONFIG_PATH, 0o600)
+    } catch {
+      return
+    }
+  }
+}
+
+function readConfigFromDisk(): AppConfig {
+  ensureConfigFile()
+  try {
+    const raw = fs.readFileSync(PATHS.CONFIG_PATH, "utf8")
+    if (!raw.trim()) {
+      fs.writeFileSync(
+        PATHS.CONFIG_PATH,
+        `${JSON.stringify(defaultConfig, null, 2)}\n`,
+        "utf8",
+      )
+      return defaultConfig
+    }
+    return JSON.parse(raw) as AppConfig
+  } catch (error) {
+    consola.error("Failed to read config file, using default config", error)
+    return defaultConfig
+  }
+}
+
+export function getConfig(): AppConfig {
+  cachedConfig ??= readConfigFromDisk()
+  return cachedConfig
+}
+
+export function getExtraPromptForModel(model: string): string {
+  const config = getConfig()
+  return config.extraPrompts?.[model] ?? ""
+}
+
+export function getSmallModel(): string {
+  const config = getConfig()
+  return config.smallModel ?? "gpt-5-mini"
+}
+
+export function getReasoningEffortForModel(
+  model: string,
+): "minimal" | "low" | "medium" | "high" {
+  const config = getConfig()
+  return config.modelReasoningEfforts?.[model] ?? "high"
+}

src/lib/logger.ts

@@ -0,0 +1,182 @@
+import consola, { type ConsolaInstance } from "consola"
+import fs from "node:fs"
+import path from "node:path"
+import util from "node:util"
+
+import { PATHS } from "./paths"
+import { state } from "./state"
+
+const LOG_RETENTION_DAYS = 7
+const LOG_RETENTION_MS = LOG_RETENTION_DAYS * 24 * 60 * 60 * 1000
+const CLEANUP_INTERVAL_MS = 24 * 60 * 60 * 1000
+const LOG_DIR = path.join(PATHS.APP_DIR, "logs")
+const FLUSH_INTERVAL_MS = 1000
+const MAX_BUFFER_SIZE = 100
+
+const logStreams = new Map<string, fs.WriteStream>()
+const logBuffers = new Map<string, Array<string>>()
+
+const ensureLogDirectory = () => {
+  if (!fs.existsSync(LOG_DIR)) {
+    fs.mkdirSync(LOG_DIR, { recursive: true })
+  }
+}
+
+const cleanupOldLogs = () => {
+  if (!fs.existsSync(LOG_DIR)) {
+    return
+  }
+
+  const now = Date.now()
+
+  for (const entry of fs.readdirSync(LOG_DIR)) {
+    const filePath = path.join(LOG_DIR, entry)
+
+    let stats: fs.Stats
+    try {
+      stats = fs.statSync(filePath)
+    } catch {
+      continue
+    }
+
+    if (!stats.isFile()) {
+      continue
+    }
+
+    if (now - stats.mtimeMs > LOG_RETENTION_MS) {
+      try {
+        fs.rmSync(filePath)
+      } catch {
+        continue
+      }
+    }
+  }
+}
+
+const formatArgs = (args: Array<unknown>) =>
+  args
+    .map((arg) =>
+      typeof arg === "string" ? arg : (
+        util.inspect(arg, { depth: null, colors: false })
+      ),
+    )
+    .join(" ")
+
+const sanitizeName = (name: string) => {
+  const normalized = name
+    .toLowerCase()
+    .replaceAll(/[^a-z0-9]+/g, "-")
+    .replaceAll(/^-+|-+$/g, "")
+
+  return normalized === "" ? "handler" : normalized
+}
+
+const getLogStream = (filePath: string): fs.WriteStream => {
+  let stream = logStreams.get(filePath)
+  if (!stream || stream.destroyed) {
+    stream = fs.createWriteStream(filePath, { flags: "a" })
+    logStreams.set(filePath, stream)
+
+    stream.on("error", (error: unknown) => {
+      console.warn("Log stream error", error)
+      logStreams.delete(filePath)
+    })
+  }
+  return stream
+}
+
+const flushBuffer = (filePath: string) => {
+  const buffer = logBuffers.get(filePath)
+  if (!buffer || buffer.length === 0) {
+    return
+  }
+
+  const stream = getLogStream(filePath)
+  const content = buffer.join("\n") + "\n"
+  stream.write(content, (error) => {
+    if (error) {
+      console.warn("Failed to write handler log", error)
+    }
+  })
+
+  logBuffers.set(filePath, [])
+}
+
+const flushAllBuffers = () => {
+  for (const filePath of logBuffers.keys()) {
+    flushBuffer(filePath)
+  }
+}
+
+const appendLine = (filePath: string, line: string) => {
+  let buffer = logBuffers.get(filePath)
+  if (!buffer) {
+    buffer = []
+    logBuffers.set(filePath, buffer)
+  }
+
+  buffer.push(line)
+
+  if (buffer.length >= MAX_BUFFER_SIZE) {
+    flushBuffer(filePath)
+  }
+}
+
+setInterval(flushAllBuffers, FLUSH_INTERVAL_MS)
+
+const cleanup = () => {
+  flushAllBuffers()
+  for (const stream of logStreams.values()) {
+    stream.end()
+  }
+  logStreams.clear()
+  logBuffers.clear()
+}
+
+process.on("exit", cleanup)
+process.on("SIGINT", () => {
+  cleanup()
+  process.exit(0)
+})
+process.on("SIGTERM", () => {
+  cleanup()
+  process.exit(0)
+})
+
+let lastCleanup = 0
+
+export const createHandlerLogger = (name: string): ConsolaInstance => {
+  ensureLogDirectory()
+
+  const sanitizedName = sanitizeName(name)
+  const instance = consola.withTag(name)
+
+  if (state.verbose) {
+    instance.level = 5
+  }
+  instance.setReporters([])
+
+  instance.addReporter({
+    log(logObj) {
+      ensureLogDirectory()
+
+      if (Date.now() - lastCleanup > CLEANUP_INTERVAL_MS) {
+        cleanupOldLogs()
+        lastCleanup = Date.now()
+      }
+
+      const date = logObj.date
+      const dateKey = date.toLocaleDateString("sv-SE")
+      const timestamp = date.toLocaleString("sv-SE", { hour12: false })
+      const filePath = path.join(LOG_DIR, `${sanitizedName}-${dateKey}.log`)
+      const message = formatArgs(logObj.args as Array<unknown>)
+      const line = `[${timestamp}] [${logObj.type}] [${logObj.tag || name}]${
+        message ? ` ${message}` : ""
+      }`
+
+      appendLine(filePath, line)
+    },
+  })
+
+  return instance
+}

src/lib/paths.ts

@@ -5,15 +5,18 @@ import path from "node:path"
 const APP_DIR = path.join(os.homedir(), ".local", "share", "copilot-api")
 
 const GITHUB_TOKEN_PATH = path.join(APP_DIR, "github_token")
+const CONFIG_PATH = path.join(APP_DIR, "config.json")
 
 export const PATHS = {
   APP_DIR,
   GITHUB_TOKEN_PATH,
+  CONFIG_PATH,
 }
 
 export async function ensurePaths(): Promise<void> {
   await fs.mkdir(PATHS.APP_DIR, { recursive: true })
   await ensureFile(PATHS.GITHUB_TOKEN_PATH)
+  await ensureFile(PATHS.CONFIG_PATH)
 }
 
 async function ensureFile(filePath: string): Promise<void> {

src/lib/state.ts

@@ -15,11 +15,13 @@ export interface State {
   // Rate limiting configuration
   rateLimitSeconds?: number
   lastRequestTimestamp?: number
+  verbose: boolean
 }
 
 export const state: State = {
   accountType: "individual",
   manualApprove: false,
   rateLimitWait: false,
   showToken: false,
+  verbose: false,
 }

src/routes/chat-completions/handler.ts

@@ -1,9 +1,9 @@
 import type { Context } from "hono"
 
-import consola from "consola"
 import { streamSSE, type SSEMessage } from "hono/streaming"
 
 import { awaitApproval } from "~/lib/approval"
+import { createHandlerLogger } from "~/lib/logger"
 import { checkRateLimit } from "~/lib/rate-limit"
 import { state } from "~/lib/state"
 import { getTokenCount } from "~/lib/tokenizer"
@@ -14,11 +14,13 @@ import {
   type ChatCompletionsPayload,
 } from "~/services/copilot/create-chat-completions"
 
+const logger = createHandlerLogger("chat-completions-handler")
+
 export async function handleCompletion(c: Context) {
   await checkRateLimit(state)
 
   let payload = await c.req.json<ChatCompletionsPayload>()
-  consola.debug("Request payload:", JSON.stringify(payload).slice(-400))
+  logger.debug("Request payload:", JSON.stringify(payload).slice(-400))
 
   // Find the selected model
   const selectedModel = state.models?.data.find(
@@ -29,12 +31,12 @@ export async function handleCompletion(c: Context) {
   try {
     if (selectedModel) {
       const tokenCount = await getTokenCount(payload, selectedModel)
-      consola.info("Current token count:", tokenCount)
+      logger.info("Current token count:", tokenCount)
     } else {
-      consola.warn("No model selected, skipping token count calculation")
+      logger.warn("No model selected, skipping token count calculation")
     }
   } catch (error) {
-    consola.warn("Failed to calculate token count:", error)
+    logger.warn("Failed to calculate token count:", error)
   }
 
   if (state.manualApprove) await awaitApproval()
@@ -44,20 +46,20 @@ export async function handleCompletion(c: Context) {
       ...payload,
       max_tokens: selectedModel?.capabilities.limits.max_output_tokens,
     }
-    consola.debug("Set max_tokens to:", JSON.stringify(payload.max_tokens))
+    logger.debug("Set max_tokens to:", JSON.stringify(payload.max_tokens))
   }
 
   const response = await createChatCompletions(payload)
 
   if (isNonStreaming(response)) {
-    consola.debug("Non-streaming response:", JSON.stringify(response))
+    logger.debug("Non-streaming response:", JSON.stringify(response))
     return c.json(response)
   }
 
-  consola.debug("Streaming response")
+  logger.debug("Streaming response")
   return streamSSE(c, async (stream) => {
     for await (const chunk of response) {
-      consola.debug("Streaming chunk:", JSON.stringify(chunk))
+      logger.debug("Streaming chunk:", JSON.stringify(chunk))
       await stream.writeSSE(chunk as SSEMessage)
     }
   })