hillclimb 0.5.4 → 0.6.0

package/dist/cli.js

@@ -9,22 +9,6 @@ import * as p6 from "@clack/prompts";
 import path7 from "path";
 import * as p3 from "@clack/prompts";
 
-// src/color.ts
-var IS_TTY = Boolean(process.stdout.isTTY);
-var NO_COLOR = process.env.FORCE_COLOR !== void 0 ? false : process.env.NO_COLOR !== void 0 || process.env.TERM === "dumb" || !IS_TTY;
-function wrap(open, close) {
-  if (NO_COLOR) return (s) => s;
-  const o = `\x1B[${open}m`;
-  const c = `\x1B[${close}m`;
-  return (s) => `${o}${s}${c}`;
-}
-var bold = wrap(1, 22);
-var dim = wrap(2, 22);
-var red = wrap(31, 39);
-var green = wrap(32, 39);
-var yellow = wrap(33, 39);
-var cyan = wrap(36, 39);
-
 // src/config.ts
 import fs from "fs";
 import os from "os";
@@ -369,18 +353,18 @@ var PlatformClient = class {
       let code;
       let message = `${method} ${relative} failed: HTTP ${res.status}`;
       try {
-        const text3 = await res.text();
-        if (text3) {
+        const text2 = await res.text();
+        if (text2) {
           try {
-            const parsed = JSON.parse(text3);
+            const parsed = JSON.parse(text2);
             if (parsed.error?.code) code = parsed.error.code;
             if (parsed.error?.message) {
               message = `${method} ${relative} failed: ${parsed.error.message}`;
             } else {
-              message = `${method} ${relative} failed: ${text3}`;
+              message = `${method} ${relative} failed: ${text2}`;
             }
           } catch {
-            message = `${method} ${relative} failed: ${text3}`;
+            message = `${method} ${relative} failed: ${text2}`;
           }
         }
       } catch {
@@ -397,17 +381,6 @@ var PlatformClient = class {
     }
     return void 0;
   }
-  async signInWithEmail(email, password2) {
-    await this.request("POST", "/api/auth/sign-in/email", {
-      email,
-      password: password2
-    });
-    if (!this.sessionCookie) {
-      throw new PlatformError(
-        "Sign-in succeeded but no session cookie was returned"
-      );
-    }
-  }
   async getBootstrap() {
     const data = await this.request("GET", "/api/v1/app/bootstrap");
     if (!data?.session?.user || !data.session.session) return null;
@@ -1483,22 +1456,6 @@ var CLAUDE_DEF = TOOLS[0];
 
 // src/commands/_shared.ts
 import * as p from "@clack/prompts";
-async function requireText(label, opts = {}) {
-  const result = opts.password ? await p.password({
-    message: label,
-    validate: (v) => v && v.length > 0 ? void 0 : "Required"
-  }) : await p.text({
-    message: label,
-    placeholder: opts.placeholder,
-    initialValue: opts.initialValue,
-    validate: (v) => v && v.trim().length > 0 ? void 0 : "Required"
-  });
-  if (p.isCancel(result)) {
-    p.cancel("Cancelled.");
-    process.exit(0);
-  }
-  return typeof result === "string" ? result.trim() : "";
-}
 async function withSpinner(startLabel, onSuccess, onFailureLabel, fn) {
   const spinner5 = p.spinner();
   spinner5.start(startLabel);
@@ -1517,6 +1474,24 @@ async function withSpinner(startLabel, onSuccess, onFailureLabel, fn) {
 // src/commands/web-auth.ts
 import { spawn } from "child_process";
 import * as p2 from "@clack/prompts";
+
+// src/color.ts
+var IS_TTY = Boolean(process.stdout.isTTY);
+var NO_COLOR = process.env.FORCE_COLOR !== void 0 ? false : process.env.NO_COLOR !== void 0 || process.env.TERM === "dumb" || !IS_TTY;
+function wrap(open, close) {
+  if (NO_COLOR) return (s) => s;
+  const o = `\x1B[${open}m`;
+  const c = `\x1B[${close}m`;
+  return (s) => `${o}${s}${c}`;
+}
+var bold = wrap(1, 22);
+var dim = wrap(2, 22);
+var red = wrap(31, 39);
+var green = wrap(32, 39);
+var yellow = wrap(33, 39);
+var cyan = wrap(36, 39);
+
+// src/commands/web-auth.ts
 function sleep(ms) {
   return new Promise((resolve) => setTimeout(resolve, ms));
 }
@@ -1531,10 +1506,6 @@ function openBrowser(url) {
   }
 }
 async function webAuth(apiBaseUrl) {
-  if (!IS_TTY) {
-    appendLog("info", "web-auth: skipped (not a TTY)");
-    return null;
-  }
   appendLog("info", `web-auth: started (apiBaseUrl=${apiBaseUrl})`);
   const client = new PlatformClient(apiBaseUrl);
   const codeResponse = await withSpinner(
@@ -1548,9 +1519,16 @@ async function webAuth(apiBaseUrl) {
     `web-auth: code prepared (expiresIn=${codeResponse.expiresIn}s, pollInterval=${codeResponse.interval}s)`
   );
   const url = codeResponse.verificationUrl;
-  openBrowser(url);
-  p2.log.info(`Opening browser to: ${cyan(url)}`);
-  p2.log.info(dim("If the browser did not open, copy the URL above."));
+  if (IS_TTY) {
+    openBrowser(url);
+    p2.log.info(`Opening browser to: ${cyan(url)}`);
+    p2.log.info(dim("If the browser did not open, copy the URL above."));
+  } else {
+    p2.log.info(
+      "To sign in, open this URL in a browser and approve the request:"
+    );
+    p2.log.info(cyan(url));
+  }
   const deadline = Date.now() + codeResponse.expiresIn * 1e3;
   const interval = codeResponse.interval * 1e3;
   const spinner5 = p2.spinner();
@@ -1674,66 +1652,24 @@ async function runInit(args = []) {
     bootstrap = await tryReuseIdentity(apiBaseUrl, client, saved);
   }
   if (!bootstrap) {
-    let method = projectSlugFlag ? "web" : "email";
-    if (IS_TTY && !projectSlugFlag) {
-      const selected = await p3.select({
-        message: "How would you like to sign in?",
-        options: [
-          {
-            value: "web",
-            label: "Web authentication (recommended)",
-            hint: "opens browser"
-          },
-          { value: "email", label: "Email and password" }
-        ],
-        initialValue: "web"
-      });
-      if (p3.isCancel(selected)) {
-        p3.cancel("Cancelled.");
-        process.exit(0);
-      }
-      method = selected;
+    const result = await webAuth(apiBaseUrl);
+    if (!result) {
+      appendLog("error", "init: web authentication failed");
+      p3.log.error("Web authentication failed.");
+      p3.outro("Aborted.");
+      process.exit(1);
     }
-    appendLog("info", `init: auth method selected (${method})`);
-    if (method === "web") {
-      const result = await webAuth(apiBaseUrl);
-      if (!result) {
-        appendLog("error", "init: web authentication failed");
-        p3.log.error("Web authentication failed.");
-        p3.outro("Aborted.");
-        process.exit(1);
+    client = result.client;
+    bootstrap = await withSpinner(
+      "Loading account...",
+      (b) => `Signed in as ${b.session.email}${b.session.role === "admin" ? " (admin)" : ""}.`,
+      "Could not load account.",
+      async () => {
+        const data = await client.getBootstrap();
+        if (!data) throw new Error("Bootstrap returned no session.");
+        return data;
       }
-      client = result.client;
-      bootstrap = await withSpinner(
-        "Loading account...",
-        (b) => `Signed in as ${b.session.email}${b.session.role === "admin" ? " (admin)" : ""}.`,
-        "Could not load account.",
-        async () => {
-          const data = await client.getBootstrap();
-          if (!data) throw new Error("Bootstrap returned no session.");
-          return data;
-        }
-      );
-    } else {
-      const email = await requireText("Email");
-      const password2 = await requireText("Password", { password: true });
-      await withSpinner(
-        "Signing in...",
-        () => "Signed in.",
-        "Sign-in failed.",
-        () => client.signInWithEmail(email, password2)
-      );
-      bootstrap = await withSpinner(
-        "Loading account...",
-        (b) => `Signed in as ${b.session.email}${b.session.role === "admin" ? " (admin)" : ""}.`,
-        "Could not load account.",
-        async () => {
-          const data = await client.getBootstrap();
-          if (!data) throw new Error("Bootstrap returned no session.");
-          return data;
-        }
-      );
-    }
+    );
   }
   let project;
   if (projectSlugFlag) {
@@ -2070,87 +2006,23 @@ async function runLogin(_args = []) {
       println();
     }
   }
-  let method = "email";
-  if (IS_TTY) {
-    const selected = await p4.select({
-      message: "How would you like to sign in?",
-      options: [
-        {
-          value: "web",
-          label: "Web authentication (recommended)",
-          hint: "opens browser"
-        },
-        { value: "email", label: "Email and password" }
-      ],
-      initialValue: "web"
-    });
-    if (p4.isCancel(selected)) {
-      p4.cancel("Cancelled.");
-      process.exit(0);
-    }
-    method = selected;
-  }
-  appendLog("info", `login: auth method selected (${method})`);
-  if (method === "web") {
-    const result = await webAuth(apiBaseUrl);
-    if (!result) {
-      appendLog("error", "login: web authentication failed");
-      p4.log.error("Web authentication failed.");
-      p4.outro("Aborted.");
-      process.exit(1);
-    }
-    await saveIdentity({
-      apiBaseUrl,
-      sessionCookie: result.sessionCookie,
-      email: result.email,
-      userId: result.userId,
-      savedAt: (/* @__PURE__ */ new Date()).toISOString()
-    });
-    appendLog("info", `login: completed via web auth (email=${result.email})`);
-    println();
-    row(CHECK, "Signed in", cyan(result.email));
-    footer("Run `npx hillclimb` in a repo to wire up auto-upload.");
-    return;
-  }
-  const email = await requireText("Email");
-  const password2 = await requireText("Password", { password: true });
-  const client = new PlatformClient(apiBaseUrl);
-  await withSpinner(
-    "Signing in...",
-    () => "Signed in.",
-    "Sign-in failed.",
-    () => client.signInWithEmail(email, password2)
-  );
-  const bootstrap = await withSpinner(
-    "Loading account...",
-    (b) => `Signed in as ${b.session.email}.`,
-    "Could not load account.",
-    async () => {
-      const data = await client.getBootstrap();
-      if (!data) throw new Error("Bootstrap returned no session.");
-      return data;
-    }
-  );
-  const cookie = client.getSessionCookie();
-  if (!cookie) {
-    appendLog("error", "login: session cookie missing after sign-in");
-    p4.log.error("Session cookie was lost during sign-in.");
+  const result = await webAuth(apiBaseUrl);
+  if (!result) {
+    appendLog("error", "login: web authentication failed");
+    p4.log.error("Web authentication failed.");
     p4.outro("Aborted.");
     process.exit(1);
   }
   await saveIdentity({
     apiBaseUrl,
-    sessionCookie: cookie,
-    email: bootstrap.session.email,
-    userId: bootstrap.session.userId,
+    sessionCookie: result.sessionCookie,
+    email: result.email,
+    userId: result.userId,
     savedAt: (/* @__PURE__ */ new Date()).toISOString()
   });
-  appendLog(
-    "info",
-    `login: completed via email/password (email=${bootstrap.session.email})`
-  );
+  appendLog("info", `login: completed via web auth (email=${result.email})`);
   println();
-  row(CHECK, "Signed in", cyan(bootstrap.session.email));
+  row(CHECK, "Signed in", cyan(result.email));
   footer("Run `npx hillclimb` in a repo to wire up auto-upload.");
 }
 
@@ -2419,6 +2291,48 @@ async function readFileContent(file) {
   }
 }
 
+// src/middleware/large-file.ts
+import { constants as bufferConstants } from "buffer";
+var MAX_STRINGIFIABLE_BYTES = bufferConstants.MAX_STRING_LENGTH;
+var largeFileThreshold = MAX_STRINGIFIABLE_BYTES;
+function getLargeFileThreshold() {
+  return largeFileThreshold;
+}
+var NEWLINE = 10;
+var OVERSIZED_LINE_PLACEHOLDER = '{"_hillclimb_omitted":"line exceeded max string length; dropped to allow redaction"}';
+function redactBufferByLine(buffer, redactLine, options = {}) {
+  const maxLineBytes = options.maxLineBytes ?? MAX_STRINGIFIABLE_BYTES;
+  const oversizedPlaceholder = options.oversizedPlaceholder ?? OVERSIZED_LINE_PLACEHOLDER;
+  const pieces = [];
+  const newlineBuf = Buffer.from([NEWLINE]);
+  let totalCount = 0;
+  let oversizedLines = 0;
+  let start = 0;
+  let first = true;
+  for (; ; ) {
+    const nl = buffer.indexOf(NEWLINE, start);
+    const end = nl === -1 ? buffer.length : nl;
+    const slice = buffer.subarray(start, end);
+    if (!first) pieces.push(newlineBuf);
+    first = false;
+    if (slice.length > maxLineBytes) {
+      oversizedLines++;
+      pieces.push(Buffer.from(oversizedPlaceholder, "utf-8"));
+    } else {
+      const { value, count } = redactLine(slice.toString("utf-8"));
+      totalCount += count;
+      pieces.push(Buffer.from(value, "utf-8"));
+    }
+    if (nl === -1) break;
+    start = nl + 1;
+  }
+  return {
+    content: Buffer.concat(pieces),
+    count: totalCount,
+    oversizedLines
+  };
+}
+
 // src/middleware/patterns.json
 var patterns_default = {
   patterns: [
@@ -10658,6 +10572,19 @@ function processFiles(files) {
 }
 
 // src/middleware/pattern-redact.ts
+function redactPatternLine(line, patterns, memo, isJsonl) {
+  if (isJsonl) {
+    if (!line.trim()) return { value: line, count: 0 };
+    try {
+      const parsed = JSON.parse(line);
+      const walked = walkAndRedactAll(parsed, patterns, memo);
+      return { value: JSON.stringify(walked.value), count: walked.count };
+    } catch {
+      return redactString(line, patterns);
+    }
+  }
+  return redactString(line, patterns);
+}
 var WORKER_COUNT = Math.min(
   os3.availableParallelism?.() ?? os3.cpus().length,
   4
@@ -10686,9 +10613,14 @@ var PatternRedactMiddleware = class {
     const tasks = [];
     const fileMap = /* @__PURE__ */ new Map();
     const passThrough = [];
+    const largeResults = /* @__PURE__ */ new Map();
     for (let i = 0; i < group.files.length; i++) {
       const file = group.files[i];
       this.stats.filesScanned++;
+      if (file.content && !file.isBinary && file.content.length > getLargeFileThreshold()) {
+        largeResults.set(i, this.redactLargeFile(file));
+        continue;
+      }
       const read = await readFileContent(file);
       if (read.kind === "binary") {
         this.stats.binaryFiles++;
@@ -10718,6 +10650,11 @@ var PatternRedactMiddleware = class {
     }
     const newFiles = [];
     for (let i = 0; i < group.files.length; i++) {
+      const large = largeResults.get(i);
+      if (large) {
+        newFiles.push(large);
+        continue;
+      }
       const pt = passThrough.find((p7) => p7.index === i);
       if (pt) {
         newFiles.push(pt.file);
@@ -10739,6 +10676,33 @@ var PatternRedactMiddleware = class {
     }
     return { ...group, files: newFiles };
   }
+  // Stream-redact an oversized file straight from its buffer, one line at a time,
+  // so we never build a string larger than a single line. Single-threaded (no
+  // worker pool) — it trades parallelism for bounded memory on the one giant file.
+  redactLargeFile(file) {
+    const buffer = file.content;
+    const isJsonl = file.absolutePath.endsWith(".jsonl");
+    const patterns = compilePatterns();
+    const memo = /* @__PURE__ */ new Map();
+    const { content, count, oversizedLines } = redactBufferByLine(
+      buffer,
+      (line) => redactPatternLine(line, patterns, memo, isJsonl)
+    );
+    if (oversizedLines > 0) {
+      appendLog(
+        "warn",
+        `pattern-redact: dropped ${oversizedLines} oversized line(s) in ${file.absolutePath} (exceeded max string length)`
+      );
+    }
+    if (count > 0) {
+      this.stats.filesRedacted++;
+      this.stats.totalRedactions += count;
+    }
+    if (count > 0 || oversizedLines > 0) {
+      return { ...file, content };
+    }
+    return file;
+  }
   async processWithWorkers(tasks) {
     const workerPath = resolveWorkerPath();
     const sorted = [...tasks].sort(
@@ -10814,6 +10778,10 @@ var RedactMiddleware = class {
     for (const file of group.files) {
       this.stats.filesScanned++;
       regex.lastIndex = 0;
+      if (file.content && !file.isBinary && file.content.length > getLargeFileThreshold()) {
+        newFiles.push(this.redactLargeFile(file, regex));
+        continue;
+      }
       const read = await readFileContent(file);
       if (read.kind === "binary") {
         this.stats.binaryFiles++;
@@ -10849,6 +10817,47 @@ var RedactMiddleware = class {
     }
     return { ...group, files: newFiles };
   }
+  // Stream-redact an oversized file straight from its buffer. Each line is
+  // redacted with the same JSON-aware (or raw) logic as redactJsonl, so the
+  // output matches the in-memory path exactly for files whose lines all fit in a
+  // string — which holds for real transcripts (largest line is a few MiB).
+  redactLargeFile(file, regex) {
+    const buffer = file.content;
+    const isJsonl = file.absolutePath.endsWith(".jsonl");
+    const { content, count, oversizedLines } = redactBufferByLine(
+      buffer,
+      (line) => this.redactLine(line, regex, isJsonl)
+    );
+    if (oversizedLines > 0) {
+      appendLog(
+        "warn",
+        `redact: dropped ${oversizedLines} oversized line(s) in ${file.absolutePath} (exceeded max string length)`
+      );
+    }
+    if (count > 0) {
+      this.stats.filesRedacted++;
+      this.stats.totalRedactions += count;
+    }
+    if (count > 0 || oversizedLines > 0) {
+      return { ...file, content };
+    }
+    return file;
+  }
+  redactLine(line, regex, isJsonl) {
+    if (isJsonl) {
+      if (!line.trim()) return { value: line, count: 0 };
+      try {
+        const parsed = JSON.parse(line);
+        const walked = walkAndRedact(parsed, regex);
+        return { value: JSON.stringify(walked.value), count: walked.count };
+      } catch {
+        const { result: result2, count: count2 } = this.redactString(line, regex);
+        return { value: result2, count: count2 };
+      }
+    }
+    const { result, count } = this.redactString(line, regex);
+    return { value: result, count };
+  }
   redactJsonl(content, regex) {
     let totalCount = 0;
     const lines = content.split("\n");
@@ -11708,9 +11717,9 @@ function extractTextReasoningToolUses(content) {
   } else if (content !== void 0 && content !== null) {
     textParts.push(stringify(content));
   }
-  const text3 = textParts.filter((p7) => p7 && p7.trim()).map((p7) => p7.trim()).join("\n\n");
+  const text2 = textParts.filter((p7) => p7 && p7.trim()).map((p7) => p7.trim()).join("\n\n");
   const reasoning = reasoningParts.filter((p7) => p7 && p7.trim()).map((p7) => p7.trim()).join("\n\n");
-  return [text3, reasoning || void 0, toolBlocks];
+  return [text2, reasoning || void 0, toolBlocks];
 }
 function buildMetrics(usage) {
   if (typeof usage !== "object" || usage === null) return void 0;
@@ -11739,8 +11748,8 @@ function formatToolResult(block, toolUseResult) {
     if (content.trim()) parts.push(content.trim());
   } else if (Array.isArray(content)) {
     for (const item of content) {
-      const text3 = stringify(item);
-      if (text3.trim()) parts.push(text3.trim());
+      const text2 = stringify(item);
+      if (text2.trim()) parts.push(text2.trim());
     }
   } else if (content !== void 0 && content !== null && content !== "") {
     parts.push(stringify(content));
@@ -11868,7 +11877,7 @@ function convertClaudeToTrajectory(jsonlContent, sessionId) {
     const eventType = event.type;
     const timestamp = event.timestamp;
     if (eventType === "assistant") {
-      const [text3, reasoning, toolBlocks] = extractTextReasoningToolUses(
+      const [text2, reasoning, toolBlocks] = extractTextReasoningToolUses(
         msg.content
       );
       const msgId = msg.id;
@@ -11892,12 +11901,12 @@ function convertClaudeToTrajectory(jsonlContent, sessionId) {
         extra.user_type = event.userType;
       extra.is_sidechain = event.isSidechain ?? false;
       const modelName = msg.model || defaultModelName;
-      if (text3 || reasoning || toolBlocks.length === 0) {
+      if (text2 || reasoning || toolBlocks.length === 0) {
         normalizedEvents.push({
           kind: "message",
           timestamp,
           role: msg.role ?? "assistant",
-          text: text3 || "",
+          text: text2 || "",
           reasoning: msg.role === "assistant" ? reasoning : void 0,
           metrics,
           extra: Object.keys(extra).length > 0 ? extra : void 0,
@@ -11937,13 +11946,13 @@ function convertClaudeToTrajectory(jsonlContent, sessionId) {
     if (eventType === "user") {
       const content = msg.content;
       if (typeof content === "string") {
-        const text3 = content.trim();
-        if (text3) {
+        const text2 = content.trim();
+        if (text2) {
           normalizedEvents.push({
             kind: "message",
             timestamp,
             role: "user",
-            text: text3,
+            text: text2,
             extra: { is_sidechain: event.isSidechain ?? false }
           });
         }
@@ -12005,13 +12014,13 @@ function convertClaudeToTrajectory(jsonlContent, sessionId) {
         continue;
       }
       if (content !== void 0 && content !== null && content !== "") {
-        const text3 = stringify(content).trim();
-        if (text3) {
+        const text2 = stringify(content).trim();
+        if (text2) {
           normalizedEvents.push({
             kind: "message",
             timestamp,
             role: "user",
-            text: text3
+            text: text2
           });
         }
       }
@@ -12147,8 +12156,8 @@ function extractMessageText(content) {
   const parts = [];
   for (const block of content) {
     if (typeof block === "object" && block !== null) {
-      const text3 = block.text;
-      if (typeof text3 === "string") parts.push(text3);
+      const text2 = block.text;
+      if (typeof text2 === "string") parts.push(text2);
     }
   }
   return parts.join("");
@@ -12280,12 +12289,12 @@ function convertCodexToTrajectory(jsonlContent, sessionId) {
     }
     if (payloadType === "message") {
       const content = payload.content;
-      const text3 = Array.isArray(content) ? extractMessageText(content) : "";
+      const text2 = Array.isArray(content) ? extractMessageText(content) : "";
       normalizedEvents.push({
         kind: "message",
         timestamp,
         role: payload.role ?? "user",
-        text: text3,
+        text: text2,
         reasoning: payload.role === "assistant" ? pendingReasoning : void 0
       });
       pendingReasoning = void 0;
@@ -12791,7 +12800,7 @@ function convertCursorToTrajectory(jsonlContent, sessionId) {
     else if (role === "user") source = "user";
     else source = "system";
     const contentParts = entry.message?.content;
-    let text3 = "";
+    let text2 = "";
     if (Array.isArray(contentParts)) {
       const textParts = [];
       for (const part of contentParts) {
@@ -12799,13 +12808,13 @@ function convertCursorToTrajectory(jsonlContent, sessionId) {
           textParts.push(part.text);
         }
       }
-      text3 = textParts.join("\n\n").trim();
+      text2 = textParts.join("\n\n").trim();
     }
-    if (!text3) continue;
+    if (!text2) continue;
     const step = {
       step_id: stepId,
       source,
-      message: text3
+      message: text2
     };
     steps.push(step);
     stepId++;
@@ -13351,10 +13360,10 @@ var NormalizeMiddleware = class {
         file.sourceName
       ))
         continue;
-      const content = file.content ? file.content.toString("utf-8") : null;
-      if (!content) continue;
       const sessionId = file.metadata?.sessionId ?? (file.sourceName === "codex" ? void 0 : path13.basename(file.absolutePath, ".jsonl"));
       try {
+        const content = file.content ? file.content.toString("utf-8") : null;
+        if (!content) continue;
         const trajectory = normalizeContent(
           file.sourceName,
           content,
@@ -13383,7 +13392,7 @@ var NormalizeMiddleware = class {
       } catch (err) {
         appendLog(
           "warn",
-          `normalize: failed for source=${file.sourceName} session=${sessionId ?? "<unknown>"} file=${file.absolutePath}: ${err instanceof Error ? err.message : String(err)}`
+          `normalize: skipped ATIF for source=${file.sourceName} session=${sessionId ?? "<unknown>"} file=${file.absolutePath}: ${err instanceof Error ? err.message : String(err)}`
         );
       }
     }
@@ -14085,7 +14094,7 @@ import { execFileSync as execFileSync3 } from "child_process";
 import path18 from "path";
 
 // src/git-traces/git-ops.ts
-import { execFileSync as execFileSync2 } from "child_process";
+import { execFileSync as execFileSync2, spawnSync } from "child_process";
 import fs12 from "fs";
 import os7 from "os";
 import path16 from "path";
@@ -14097,11 +14106,155 @@ var EXEC_OPTS = {
 };
 var MAX_ERROR_OUTPUT_CHARS = 2e3;
 var MAX_SNAPSHOT_FILE_BYTES = 10 * 1024 * 1024;
-var EXCLUDED_SNAPSHOT_EXTENSIONS = /* @__PURE__ */ new Set([".pdf"]);
+var MAX_BINARY_SNAPSHOT_FILE_BYTES = 1024 * 1024;
+var BINARY_SNIFF_BYTES = 8e3;
+var EXCLUDED_SNAPSHOT_EXTENSIONS = /* @__PURE__ */ new Set([
+  // Documents
+  ".pdf",
+  // Raster images
+  ".png",
+  ".jpg",
+  ".jpeg",
+  ".gif",
+  ".bmp",
+  ".tiff",
+  ".tif",
+  ".webp",
+  ".ico",
+  ".heic",
+  ".heif",
+  ".avif",
+  // Video
+  ".mp4",
+  ".mov",
+  ".avi",
+  ".mkv",
+  ".webm",
+  ".m4v",
+  ".mpg",
+  ".mpeg",
+  ".wmv",
+  ".flv",
+  // Audio
+  ".mp3",
+  ".wav",
+  ".flac",
+  ".ogg",
+  ".m4a",
+  ".aac",
+  ".aiff",
+  // Archives / compressed
+  ".zip",
+  ".tar",
+  ".gz",
+  ".tgz",
+  ".bz2",
+  ".tbz2",
+  ".xz",
+  ".7z",
+  ".rar",
+  ".zst",
+  ".lz4",
+  // Fonts
+  ".woff",
+  ".woff2",
+  ".ttf",
+  ".otf",
+  ".eot",
+  // Compiled / native artifacts
+  ".so",
+  ".dylib",
+  ".dll",
+  ".a",
+  ".o",
+  ".lib",
+  ".class",
+  ".jar",
+  ".wasm",
+  ".exe",
+  ".pyc",
+  ".pyd",
+  ".node",
+  // Databases
+  ".sqlite",
+  ".sqlite3",
+  ".db",
+  ".mdb",
+  // Office / design binaries
+  ".doc",
+  ".docx",
+  ".xls",
+  ".xlsx",
+  ".ppt",
+  ".pptx",
+  ".odt",
+  ".ods",
+  ".odp",
+  ".psd",
+  ".ai",
+  ".sketch",
+  ".fig",
+  ".blend"
+]);
+var EXCLUDED_SNAPSHOT_BASENAMES = /* @__PURE__ */ new Set([
+  ".DS_Store",
+  "Thumbs.db",
+  "desktop.ini"
+]);
 var EMPTY_TREE_SHA = "4b825dc642cb6eb9a060e54bf8d69288fbee4904";
-function hasExcludedSnapshotExtension(filePath) {
+function isExcludedSnapshotPath(filePath) {
+  if (EXCLUDED_SNAPSHOT_BASENAMES.has(path16.basename(filePath))) return true;
   return EXCLUDED_SNAPSHOT_EXTENSIONS.has(path16.extname(filePath).toLowerCase());
 }
+function isBinaryBuffer(buffer) {
+  return buffer.includes(0);
+}
+function readTreeBlobHead(repoRoot, sha) {
+  try {
+    const { stdout } = spawnSync("git", ["cat-file", "blob", sha], {
+      cwd: repoRoot,
+      timeout: GIT_COMMAND_TIMEOUT_MS,
+      maxBuffer: BINARY_SNIFF_BYTES
+    });
+    return stdout?.length ? stdout.subarray(0, BINARY_SNIFF_BYTES) : null;
+  } catch {
+    return null;
+  }
+}
+function readWorkingFileHead(absPath) {
+  let fd = null;
+  try {
+    fd = fs12.openSync(absPath, "r");
+    const buffer = Buffer.alloc(BINARY_SNIFF_BYTES);
+    const bytesRead = fs12.readSync(fd, buffer, 0, BINARY_SNIFF_BYTES, 0);
+    return buffer.subarray(0, bytesRead);
+  } catch {
+    return null;
+  } finally {
+    if (fd !== null) {
+      try {
+        fs12.closeSync(fd);
+      } catch {
+      }
+    }
+  }
+}
+function classifyOmission(filePath, sizeBytes, readHead) {
+  if (isExcludedSnapshotPath(filePath)) return "excluded-extension";
+  if (sizeBytes > MAX_SNAPSHOT_FILE_BYTES) return "file-over-limit";
+  if (sizeBytes > MAX_BINARY_SNAPSHOT_FILE_BYTES) {
+    const head = readHead();
+    if (!head) {
+      appendLog(
+        "warn",
+        `git-traces: binary sniff failed for ${filePath} (${sizeBytes} bytes); keeping it in the snapshot`
+      );
+      return null;
+    }
+    if (isBinaryBuffer(head)) return "binary-over-limit";
+  }
+  return null;
+}
 function quoteGitArg(arg) {
   if (/^[A-Za-z0-9_./:=@%+,-]+$/.test(arg)) return arg;
   return JSON.stringify(arg);
@@ -14205,7 +14358,7 @@ function recordOmittedSnapshotFile(omittedFiles, file, options = {}) {
   omittedFiles?.push(file);
   if (options.log === false) return;
   const action = file.tracked ? "omitting tracked" : "skipping untracked";
-  const reason = file.reason === "file-over-limit" ? `${file.sizeBytes} bytes > ${MAX_SNAPSHOT_FILE_BYTES} limit` : "excluded extension";
+  const reason = file.reason === "file-over-limit" ? `${file.sizeBytes} bytes > ${MAX_SNAPSHOT_FILE_BYTES} limit` : file.reason === "binary-over-limit" ? `binary ${file.sizeBytes} bytes > ${MAX_BINARY_SNAPSHOT_FILE_BYTES} binary limit` : "excluded extension";
   appendLog("warn", `git-traces: ${action} ${file.path} (${reason})`);
 }
 function parseLsTreeLongZ(output) {
@@ -14228,7 +14381,11 @@ function listOmittedTreeFiles(repoRoot, treeSha, options = {}) {
   const output = gitBuffer(repoRoot, ["ls-tree", "-r", "-l", "-z", treeSha]);
   const omitted = [];
   for (const entry of parseLsTreeLongZ(output)) {
-    const reason = hasExcludedSnapshotExtension(entry.path) ? "excluded-extension" : entry.sizeBytes > MAX_SNAPSHOT_FILE_BYTES ? "file-over-limit" : null;
+    const reason = classifyOmission(
+      entry.path,
+      entry.sizeBytes,
+      () => readTreeBlobHead(repoRoot, entry.sha)
+    );
     if (!reason) continue;
     const file = {
       path: entry.path,
@@ -14286,8 +14443,13 @@ function buildUntrackedTree(repoRoot, omittedFiles) {
   for (const relPath of list.split("\0")) {
     if (!relPath) continue;
     try {
-      const stat = fs12.lstatSync(path16.join(repoRoot, relPath));
-      const reason = hasExcludedSnapshotExtension(relPath) ? "excluded-extension" : stat.size > MAX_SNAPSHOT_FILE_BYTES ? "file-over-limit" : null;
+      const absPath = path16.join(repoRoot, relPath);
+      const stat = fs12.lstatSync(absPath);
+      const reason = classifyOmission(
+        relPath,
+        stat.size,
+        () => readWorkingFileHead(absPath)
+      );
       if (reason) {
         recordOmittedSnapshotFile(omittedFiles, {
           path: relPath,

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "hillclimb",
-  "version": "0.5.4",
+  "version": "0.6.0",
   "description": "Extract and export AI coding tool logs grouped by repo",
   "license": "MIT",
   "type": "module",