vidistill 0.5.4 → 0.6.0

package/README.md

@@ -104,6 +104,35 @@ vidistill rename-speakers ./vidistill-output/my-meeting/ --rename "Steven Kang"
 vidistill rename-speakers ./vidistill-output/my-meeting/ --merge "K Iphone" "Kristian"
 ```
 
+## MCP Server
+
+vidistill can run as an MCP server, letting AI coding tools (Claude Code, Cursor, etc.) analyze videos and read output directly.
+
+```bash
+vidistill mcp
+```
+
+To configure in Claude Code, add to `~/.claude/claude_code_config.json`:
+
+```json
+{
+  "mcpServers": {
+    "vidistill": {
+      "command": "npx",
+      "args": ["vidistill", "mcp"]
+    }
+  }
+}
+```
+
+**Tools exposed:**
+
+- `analyze_video` — run the full pipeline on a URL or file, returns output dir + summary
+- `get_transcript` — read transcript from an existing output dir, with optional time range filtering
+- `get_code` — read extracted code files from an existing output dir
+
+Requires `GEMINI_API_KEY` set as environment variable or in `~/.vidistill/config.json`.
+
 ## How It Works
 
 Supported video formats: MP4, MOV, WebM, MKV, AVI, MPEG, FLV, WMV, 3GPP. Supported audio formats: MP3, AAC, WAV, FLAC, OGG, M4A.

package/dist/index.js

@@ -648,22 +648,22 @@ function normalizeYouTubeUrl(url) {
   return `https://www.youtube.com/watch?v=${id}`;
 }
 function fetchYtDlpDuration(url) {
-  return new Promise((resolve2) => {
+  return new Promise((resolve3) => {
     execFile("yt-dlp", ["--dump-json", "--no-download", url], { timeout: 15e3 }, (err, stdout) => {
       if (err) {
-        resolve2(void 0);
+        resolve3(void 0);
         return;
       }
       try {
         const data = JSON.parse(stdout);
         const dur = data["duration"];
         if (typeof dur === "number" && dur > 0) {
-          resolve2(dur);
+          resolve3(dur);
         } else {
-          resolve2(void 0);
+          resolve3(void 0);
         }
       } catch {
-        resolve2(void 0);
+        resolve3(void 0);
       }
     });
   });
@@ -906,7 +906,7 @@ var _require = createRequire(import.meta.url);
 var ffmpeg = _require("fluent-ffmpeg");
 var BYTES_PER_SECOND = 5e5;
 function ffprobeAsync(filePath) {
-  return new Promise((resolve2, reject) => {
+  return new Promise((resolve3, reject) => {
     ffmpeg.ffprobe(filePath, (err, data) => {
       if (err) {
         reject(err);
@@ -917,7 +917,7 @@ function ffprobeAsync(filePath) {
         reject(new Error("ffprobe returned no duration"));
         return;
       }
-      resolve2(duration);
+      resolve3(duration);
     });
   });
 }
@@ -5274,17 +5274,228 @@ async function runDistill(args) {
 }
 
 // src/commands/mcp.ts
-import { log as log8 } from "@clack/prompts";
+import { resolve as resolve2, join as join5, basename as basename4, extname as extname3 } from "path";
+import { existsSync as existsSync4 } from "fs";
+import { readdir as readdir2, readFile as readFile3 } from "fs/promises";
+var DEFAULT_OUTPUT = "./vidistill-output/";
+async function resolveApiKeyNonInteractive() {
+  const envKey = process.env["GEMINI_API_KEY"];
+  if (envKey && envKey.trim().length > 0) return envKey.trim();
+  const config = await loadConfig();
+  if (config?.apiKey && config.apiKey.trim().length > 0) return config.apiKey.trim();
+  throw new Error("GEMINI_API_KEY not set. Set it as an environment variable or in ~/.vidistill/config.json");
+}
+async function analyzeVideo(input, context, lang) {
+  const apiKey = await resolveApiKeyNonInteractive();
+  const resolved = resolveInput(input);
+  const client = new GeminiClient(apiKey);
+  let fileUri;
+  let mimeType;
+  let duration;
+  let videoTitle;
+  if (resolved.type === "youtube") {
+    const result = await handleYouTube(resolved.value, client);
+    fileUri = result.fileUri;
+    mimeType = result.mimeType;
+    try {
+      duration = await detectDuration({
+        ytDlpDuration: result.duration,
+        geminiDuration: result.duration
+      });
+    } catch (err) {
+      process.stderr.write(`Duration detection failed, using 600s fallback: ${err instanceof Error ? err.message : String(err)}
+`);
+      duration = 600;
+    }
+    const videoId = extractVideoId(resolved.value);
+    videoTitle = videoId != null ? `youtube-${videoId}` : resolved.value;
+  } else {
+    const result = await handleLocalFile(resolved.value, client);
+    fileUri = result.fileUri;
+    mimeType = result.mimeType;
+    duration = await detectDuration({
+      filePath: resolved.value,
+      geminiDuration: result.duration
+    });
+    videoTitle = basename4(resolved.value, extname3(resolved.value));
+  }
+  const model = MODELS.flash;
+  const outputDir = resolve2(DEFAULT_OUTPUT);
+  const slug = slugify(videoTitle);
+  const finalOutputDir = `${outputDir}/${slug}`;
+  const rateLimiter = new RateLimiter();
+  const pipelineResult = await runPipeline({
+    client,
+    fileUri,
+    mimeType,
+    duration,
+    model,
+    context,
+    lang,
+    rateLimiter
+  });
+  await generateOutput({
+    pipelineResult,
+    outputDir,
+    videoTitle,
+    source: input,
+    duration,
+    model,
+    processingTimeMs: 0
+  });
+  let summary = "Analysis complete.";
+  const synthesisPath = join5(finalOutputDir, "raw", "synthesis.json");
+  const synthesis = await readJsonFile(synthesisPath);
+  if (synthesis?.overview) {
+    summary = synthesis.overview;
+  }
+  return { outputDir: finalOutputDir, summary };
+}
+async function getTranscript(outputDir, startTime, endTime) {
+  const absDir = resolve2(outputDir);
+  if (!existsSync4(absDir)) {
+    throw new Error("Not a vidistill output directory");
+  }
+  const rawDir = join5(absDir, "raw");
+  if (!existsSync4(rawDir)) {
+    throw new Error("No extracted data found");
+  }
+  const files = await readdir2(rawDir);
+  const pass1Files = files.filter((f) => /^pass1-seg\d+\.json$/.test(f)).sort();
+  if (pass1Files.length === 0) {
+    throw new Error("No extracted data found");
+  }
+  const lines = [];
+  for (const file of pass1Files) {
+    const data = await readJsonFile(join5(rawDir, file));
+    if (data?.transcript_entries == null) continue;
+    for (const entry of data.transcript_entries) {
+      if (startTime != null || endTime != null) {
+        const ts = parseTimestamp(entry.timestamp);
+        if (startTime != null && ts < startTime) continue;
+        if (endTime != null && ts > endTime) continue;
+      }
+      const speaker = entry.speaker ? `${entry.speaker}: ` : "";
+      lines.push(`[${entry.timestamp}] ${speaker}${entry.text}`);
+    }
+  }
+  return lines.join("\n");
+}
+async function getCode(outputDir) {
+  const absDir = resolve2(outputDir);
+  if (!existsSync4(absDir)) {
+    throw new Error("Not a vidistill output directory");
+  }
+  const codeDir = join5(absDir, "code");
+  if (!existsSync4(codeDir)) {
+    return [];
+  }
+  const files = await readdir2(codeDir);
+  const results = [];
+  for (const file of files) {
+    const content = await readFile3(join5(codeDir, file), "utf8");
+    results.push({ filename: file, content });
+  }
+  return results;
+}
 async function run(_args) {
-  log8.info("Not implemented yet.");
+  const { McpServer } = await import("@modelcontextprotocol/sdk/server/mcp.js");
+  const { StdioServerTransport } = await import("@modelcontextprotocol/sdk/server/stdio.js");
+  const z = await import("zod");
+  const server = new McpServer({
+    name: "vidistill",
+    version: "1.0.0"
+  });
+  server.registerTool(
+    "analyze_video",
+    {
+      title: "Analyze Video",
+      description: "Run the full vidistill pipeline on a video URL or local file. Returns the output directory and a summary.",
+      inputSchema: z.object({
+        input: z.string().describe("YouTube URL or local file path"),
+        context: z.string().optional().describe('Optional context about the video (e.g. "CS lecture", "product demo")'),
+        lang: z.string().optional().describe("Output language")
+      })
+    },
+    async ({ input, context, lang }) => {
+      try {
+        const result = await analyzeVideo(input, context, lang);
+        return {
+          content: [{ type: "text", text: JSON.stringify(result) }]
+        };
+      } catch (err) {
+        const message = err instanceof Error ? err.message : String(err);
+        return {
+          content: [{ type: "text", text: message }],
+          isError: true
+        };
+      }
+    }
+  );
+  server.registerTool(
+    "get_transcript",
+    {
+      title: "Get Transcript",
+      description: "Read transcript from an existing vidistill output directory. Optionally filter by time range.",
+      inputSchema: z.object({
+        outputDir: z.string().describe("Path to a vidistill output directory"),
+        startTime: z.number().optional().describe("Start time in seconds to filter from"),
+        endTime: z.number().optional().describe("End time in seconds to filter to")
+      })
+    },
+    async ({ outputDir, startTime, endTime }) => {
+      try {
+        const text4 = await getTranscript(outputDir, startTime, endTime);
+        return {
+          content: [{ type: "text", text: text4 }]
+        };
+      } catch (err) {
+        const message = err instanceof Error ? err.message : String(err);
+        return {
+          content: [{ type: "text", text: message }],
+          isError: true
+        };
+      }
+    }
+  );
+  server.registerTool(
+    "get_code",
+    {
+      title: "Get Code",
+      description: "Read code files from an existing vidistill output directory.",
+      inputSchema: z.object({
+        outputDir: z.string().describe("Path to a vidistill output directory")
+      })
+    },
+    async ({ outputDir }) => {
+      try {
+        const files = await getCode(outputDir);
+        return {
+          content: [{ type: "text", text: JSON.stringify(files) }]
+        };
+      } catch (err) {
+        const message = err instanceof Error ? err.message : String(err);
+        return {
+          content: [{ type: "text", text: message }],
+          isError: true
+        };
+      }
+    }
+  );
+  const transport = new StdioServerTransport();
+  await server.connect(transport);
+  process.on("SIGINT", async () => {
+    await server.close();
+    process.exit(0);
+  });
 }
 
 // src/commands/rename-speakers.ts
-import { join as join5 } from "path";
-import { log as log10, text as text3, isCancel as isCancel3, cancel as cancel4 } from "@clack/prompts";
+import { join as join6 } from "path";
+import { log as log9, text as text3, isCancel as isCancel3, cancel as cancel4 } from "@clack/prompts";
 
 // src/cli/speaker-naming.ts
-import { log as log9, text as text2, confirm as confirm2, isCancel as isCancel2, cancel as cancel3 } from "@clack/prompts";
+import { log as log8, text as text2, confirm as confirm2, isCancel as isCancel2, cancel as cancel3 } from "@clack/prompts";
 async function detectAndPromptMerges(mapping) {
   const byName = /* @__PURE__ */ new Map();
   for (const [label, name] of Object.entries(mapping)) {
@@ -5323,7 +5534,7 @@ async function detectAndPromptMerges(mapping) {
 async function collectSpeakersFromRaw(rawDir) {
   const speakerEntries = /* @__PURE__ */ new Map();
   for (let n = 0; n < 1e3; n++) {
-    const pass1 = await readJsonFile(join5(rawDir, `pass1-seg${n}.json`));
+    const pass1 = await readJsonFile(join6(rawDir, `pass1-seg${n}.json`));
     if (pass1 == null) break;
     for (const info of pass1.speaker_summary) {
       if (!info.speaker_id) continue;
@@ -5440,22 +5651,22 @@ function formatNameList(names) {
   return names.map((n) => `"${n.replace(/"/g, '\\"')}"`).join(", ");
 }
 async function runList(outputDir) {
-  const metadataPath = join5(outputDir, "metadata.json");
+  const metadataPath = join6(outputDir, "metadata.json");
   const metadata = await readJsonFile(metadataPath);
   if (metadata == null) {
-    log10.error("Not a vidistill output directory");
+    log9.error("Not a vidistill output directory");
     return;
   }
-  const rawDir = join5(outputDir, "raw");
+  const rawDir = join6(outputDir, "raw");
   const speakers = await collectSpeakersFromRaw(rawDir);
   const speakerMapping = metadata.speakerMapping ?? {};
   if (speakers.length === 0 && Object.keys(speakerMapping).length === 0) {
-    log10.info("No speakers found.");
+    log9.info("No speakers found.");
     return;
   }
   const groups = groupSpeakersByExistingMapping(speakers, speakerMapping);
   if (groups.length === 0) {
-    log10.info("No speakers found.");
+    log9.info("No speakers found.");
     return;
   }
   const lines = groups.map((group, idx) => {
@@ -5464,21 +5675,21 @@ async function runList(outputDir) {
     const labelsStr = group.labels.join(", ");
     return `${String(num)}. ${displayName} (${labelsStr}, ${String(group.totalEntries)} entries)`;
   });
-  log10.info(lines.join("\n"));
+  log9.info(lines.join("\n"));
 }
 async function runRename(outputDir, oldName, newName) {
   if (newName.trim().length === 0) {
-    log10.error("New name cannot be empty. Use the interactive prompt to clear a mapping.");
+    log9.error("New name cannot be empty. Use the interactive prompt to clear a mapping.");
     return;
   }
-  const metadataPath = join5(outputDir, "metadata.json");
+  const metadataPath = join6(outputDir, "metadata.json");
   const metadata = await readJsonFile(metadataPath);
   if (metadata == null) {
-    log10.error("Not a vidistill output directory");
+    log9.error("Not a vidistill output directory");
     return;
   }
   const speakerMapping = { ...metadata.speakerMapping ?? {} };
-  const rawDir = join5(outputDir, "raw");
+  const rawDir = join6(outputDir, "raw");
   const speakers = await collectSpeakersFromRaw(rawDir);
   const matchingKeys = [];
   const directKey = speakers.find((s) => s.label === oldName);
@@ -5495,18 +5706,18 @@ async function runRename(outputDir, oldName, newName) {
     const currentNames = Object.values(speakerMapping);
     const unmappedLabels = speakers.filter((s) => speakerMapping[s.label] == null).map((s) => s.label);
     const allNames = [.../* @__PURE__ */ new Set([...currentNames, ...unmappedLabels])];
-    log10.error(`No speaker named "${oldName}" found. Current speakers: ${formatNameList(allNames)}`);
+    log9.error(`No speaker named "${oldName}" found. Current speakers: ${formatNameList(allNames)}`);
     return;
   }
   if (matchingKeys.length > 1) {
-    log10.error(
+    log9.error(
       `Multiple speakers named "${oldName}" (${matchingKeys.join(", ")}). Use SPEAKER_XX label to specify which one.`
     );
     return;
   }
   const key = matchingKeys[0];
   speakerMapping[key] = newName;
-  log10.info("Re-rendering output files with updated speaker names...");
+  log9.info("Re-rendering output files with updated speaker names...");
   const result = await reRenderWithSpeakerMapping({
     outputDir,
     speakerMapping,
@@ -5514,20 +5725,20 @@ async function runRename(outputDir, oldName, newName) {
   });
   if (result.errors.length > 0) {
     for (const err of result.errors) {
-      log10.error(err);
+      log9.error(err);
     }
   }
-  log10.info(`Done. ${String(result.filesGenerated.length)} file${result.filesGenerated.length === 1 ? "" : "s"} updated.`);
+  log9.info(`Done. ${String(result.filesGenerated.length)} file${result.filesGenerated.length === 1 ? "" : "s"} updated.`);
 }
 async function runMerge(outputDir, sourceName, targetName) {
-  const metadataPath = join5(outputDir, "metadata.json");
+  const metadataPath = join6(outputDir, "metadata.json");
   const metadata = await readJsonFile(metadataPath);
   if (metadata == null) {
-    log10.error("Not a vidistill output directory");
+    log9.error("Not a vidistill output directory");
     return;
   }
   const speakerMapping = { ...metadata.speakerMapping ?? {} };
-  const rawDir = join5(outputDir, "raw");
+  const rawDir = join6(outputDir, "raw");
   const speakers = await collectSpeakersFromRaw(rawDir);
   function findKeys(name) {
     const directKey = speakers.find((s) => s.label === name);
@@ -5552,19 +5763,19 @@ async function runMerge(outputDir, sourceName, targetName) {
   const targetKeys = findKeys(targetName);
   if (sourceKeys.length === 0) {
     const currentNames = buildCurrentNames(speakers, speakerMapping);
-    log10.error(`No speaker named "${sourceName}" found. Current speakers: ${formatNameList(currentNames)}`);
+    log9.error(`No speaker named "${sourceName}" found. Current speakers: ${formatNameList(currentNames)}`);
     return;
   }
   if (targetKeys.length === 0) {
     const currentNames = buildCurrentNames(speakers, speakerMapping);
-    log10.error(`No speaker named "${targetName}" found. Current speakers: ${formatNameList(currentNames)}`);
+    log9.error(`No speaker named "${targetName}" found. Current speakers: ${formatNameList(currentNames)}`);
     return;
   }
   const resolvedTargetName = speakerMapping[targetKeys[0]] ?? targetName;
   for (const key of sourceKeys) {
     speakerMapping[key] = resolvedTargetName;
   }
-  log10.info("Re-rendering output files with updated speaker names...");
+  log9.info("Re-rendering output files with updated speaker names...");
   const result = await reRenderWithSpeakerMapping({
     outputDir,
     speakerMapping,
@@ -5572,10 +5783,10 @@ async function runMerge(outputDir, sourceName, targetName) {
   });
   if (result.errors.length > 0) {
     for (const err of result.errors) {
-      log10.error(err);
+      log9.error(err);
     }
   }
-  log10.info(`Done. ${String(result.filesGenerated.length)} file${result.filesGenerated.length === 1 ? "" : "s"} updated.`);
+  log9.info(`Done. ${String(result.filesGenerated.length)} file${result.filesGenerated.length === 1 ? "" : "s"} updated.`);
 }
 function buildCurrentNames(speakers, speakerMapping) {
   const names = /* @__PURE__ */ new Set();
@@ -5592,11 +5803,11 @@ function buildCurrentNames(speakers, speakerMapping) {
 async function run2(args) {
   const { outputDir, list, rename, merge, error } = parseArgs(args);
   if (error != null) {
-    log10.error(error);
+    log9.error(error);
     return;
   }
   if (outputDir == null || outputDir.trim() === "") {
-    log10.error('Usage: vidistill rename-speakers <output-dir> [--list] [--rename "old" "new"] [--merge "source" "target"]');
+    log9.error('Usage: vidistill rename-speakers <output-dir> [--list] [--rename "old" "new"] [--merge "source" "target"]');
     return;
   }
   if (list) {
@@ -5611,25 +5822,25 @@ async function run2(args) {
     await runMerge(outputDir, merge[0], merge[1]);
     return;
   }
-  const metadataPath = join5(outputDir, "metadata.json");
+  const metadataPath = join6(outputDir, "metadata.json");
   const metadata = await readJsonFile(metadataPath);
   if (metadata == null) {
-    log10.error("Not a vidistill output directory");
+    log9.error("Not a vidistill output directory");
     return;
   }
-  const rawDir = join5(outputDir, "raw");
-  const peopleExtraction = await readJsonFile(join5(rawDir, "pass3b-people.json"));
+  const rawDir = join6(outputDir, "raw");
+  const peopleExtraction = await readJsonFile(join6(rawDir, "pass3b-people.json"));
   if (peopleExtraction == null) {
-    log10.info("No speakers detected in this video");
+    log9.info("No speakers detected in this video");
     return;
   }
   const speakers = await collectSpeakersFromRaw(rawDir);
   if (speakers.length === 0) {
-    log10.info("No speakers detected in this video");
+    log9.info("No speakers detected in this video");
     return;
   }
   const existingMapping = metadata.speakerMapping ?? {};
-  log10.info(
+  log9.info(
     `${String(speakers.length)} speaker${speakers.length === 1 ? "" : "s"} found. Enter names (or press Enter to keep current).`
   );
   const groups = groupSpeakersByExistingMapping(speakers, existingMapping);
@@ -5676,7 +5887,7 @@ async function run2(args) {
     return;
   }
   const { mapping: finalMapping, declinedMerges } = mergeResult;
-  log10.info("Re-rendering output files with updated speaker names...");
+  log9.info("Re-rendering output files with updated speaker names...");
   const result = await reRenderWithSpeakerMapping({
     outputDir,
     speakerMapping: finalMapping,
@@ -5684,15 +5895,15 @@ async function run2(args) {
   });
   if (result.errors.length > 0) {
     for (const err of result.errors) {
-      log10.error(err);
+      log9.error(err);
     }
   }
-  log10.info(`Done. ${String(result.filesGenerated.length)} file${result.filesGenerated.length === 1 ? "" : "s"} updated.`);
+  log9.info(`Done. ${String(result.filesGenerated.length)} file${result.filesGenerated.length === 1 ? "" : "s"} updated.`);
 }
 
 // src/cli/index.ts
-var version = "0.5.4";
-var DEFAULT_OUTPUT = "./vidistill-output/";
+var version = "0.6.0";
+var DEFAULT_OUTPUT2 = "./vidistill-output/";
 var SUBCOMMANDS = {
   mcp: run,
   "rename-speakers": run2
@@ -5718,9 +5929,9 @@ Commands: ${Object.keys(SUBCOMMANDS).join(", ")}`
     },
     output: {
       type: "string",
-      description: `Output directory for generated notes (default: ${DEFAULT_OUTPUT})`,
+      description: `Output directory for generated notes (default: ${DEFAULT_OUTPUT2})`,
       alias: "o",
-      default: DEFAULT_OUTPUT
+      default: DEFAULT_OUTPUT2
     },
     lang: {
       type: "string",
@@ -5744,11 +5955,11 @@ Commands: ${Object.keys(SUBCOMMANDS).join(", ")}`
         lang: args.lang
       });
     } catch (err) {
-      const { log: log11 } = await import("@clack/prompts");
+      const { log: log10 } = await import("@clack/prompts");
       const { default: pc4 } = await import("picocolors");
       const raw = err instanceof Error ? err.message : String(err);
       const message = raw.split("\n")[0].slice(0, 200);
-      log11.error(pc4.red(message));
+      log10.error(pc4.red(message));
       process.exit(1);
     }
   }

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "vidistill",
-  "version": "0.5.4",
+  "version": "0.6.0",
   "description": "Video intelligence distiller — extract structured notes, transcripts, and insights from any video using Gemini",
   "type": "module",
   "license": "MIT",
@@ -31,11 +31,13 @@
   "dependencies": {
     "@clack/prompts": "1.0.1",
     "@google/genai": "^1.40.0",
+    "@modelcontextprotocol/sdk": "^1.27.1",
     "citty": "^0.1.6",
     "figlet": "^1.8.0",
     "fluent-ffmpeg": "^2.1.3",
     "picocolors": "^1.1.1",
-    "ytdlp-nodejs": "^2.2.0"
+    "ytdlp-nodejs": "^2.2.0",
+    "zod": "^4.3.6"
   },
   "devDependencies": {
     "@types/figlet": "^1.7.0",