mcp-scraper 0.5.0 → 0.7.0

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (52) hide show
  1. package/dist/bin/api-server.cjs +675 -214
  2. package/dist/bin/api-server.cjs.map +1 -1
  3. package/dist/bin/api-server.js +2 -2
  4. package/dist/bin/mcp-scraper-cli.cjs +1 -1
  5. package/dist/bin/mcp-scraper-cli.cjs.map +1 -1
  6. package/dist/bin/mcp-scraper-cli.js +1 -1
  7. package/dist/bin/mcp-scraper-install.cjs +1 -1
  8. package/dist/bin/mcp-scraper-install.cjs.map +1 -1
  9. package/dist/bin/mcp-scraper-install.js +1 -1
  10. package/dist/bin/mcp-stdio-server.cjs +322 -11
  11. package/dist/bin/mcp-stdio-server.cjs.map +1 -1
  12. package/dist/bin/mcp-stdio-server.js +3 -3
  13. package/dist/bin/paa-harvest.cjs +166 -74
  14. package/dist/bin/paa-harvest.cjs.map +1 -1
  15. package/dist/bin/paa-harvest.js +1 -1
  16. package/dist/chunk-EGWJ74EX.js +7 -0
  17. package/dist/chunk-EGWJ74EX.js.map +1 -0
  18. package/dist/{chunk-APJO2XV5.js → chunk-HL33CGJF.js} +3 -3
  19. package/dist/{chunk-APJO2XV5.js.map → chunk-HL33CGJF.js.map} +1 -1
  20. package/dist/{chunk-AUCXKRRH.js → chunk-RMPPYKUV.js} +167 -75
  21. package/dist/chunk-RMPPYKUV.js.map +1 -0
  22. package/dist/{chunk-BL7BBSYF.js → chunk-RUGJE5EB.js} +2 -2
  23. package/dist/{chunk-J4T5OSCF.js → chunk-YGRZU7IR.js} +322 -11
  24. package/dist/chunk-YGRZU7IR.js.map +1 -0
  25. package/dist/index.cjs +166 -74
  26. package/dist/index.cjs.map +1 -1
  27. package/dist/index.js +1 -1
  28. package/dist/{server-Z4MEISH5.js → server-JNY4XPZE.js} +138 -81
  29. package/dist/server-JNY4XPZE.js.map +1 -0
  30. package/dist/{site-extract-repository-U476J44K.js → site-extract-repository-NVSZH35Y.js} +3 -3
  31. package/dist/{worker-KJZ3ZN2N.js → worker-JQTS437L.js} +4 -4
  32. package/dist/worker-JQTS437L.js.map +1 -0
  33. package/package.json +2 -2
  34. package/dist/chunk-3RTPEQJC.js +0 -7
  35. package/dist/chunk-3RTPEQJC.js.map +0 -1
  36. package/dist/chunk-AUCXKRRH.js.map +0 -1
  37. package/dist/chunk-J4T5OSCF.js.map +0 -1
  38. package/dist/server-Z4MEISH5.js.map +0 -1
  39. package/dist/worker-KJZ3ZN2N.js.map +0 -1
  40. package/docs/spec-browser-routing.md +0 -108
  41. package/docs/spec-kernel-computer-controls.md +0 -166
  42. package/docs/spec-mcp-tools.md +0 -146
  43. package/docs/spec-server-instructions.md +0 -94
  44. package/docs/specs/agent-ready-seo-packet-spec.md +0 -237
  45. package/docs/specs/audit-visual-demo.md +0 -309
  46. package/docs/specs/cli-agent-wiring-spec.md +0 -206
  47. package/docs/specs/local-competitive-audit-spec.md +0 -311
  48. package/docs/specs/scheduled-workflows-api-spec.md +0 -304
  49. package/docs/specs/seo-cli-growth-roadmap-spec.md +0 -181
  50. package/docs/specs/seo-workflow-runner-and-reports-spec.md +0 -241
  51. /package/dist/{chunk-BL7BBSYF.js.map → chunk-RUGJE5EB.js.map} +0 -0
  52. /package/dist/{site-extract-repository-U476J44K.js.map → site-extract-repository-NVSZH35Y.js.map} +0 -0
@@ -6372,8 +6372,8 @@ async function downloadAsset(url, destDir, filename) {
6372
6372
  }
6373
6373
  const writer = (0, import_node_fs.createWriteStream)(dest);
6374
6374
  await (0, import_promises2.pipeline)(import_node_stream.Readable.fromWeb(res.body), writer);
6375
- const { statSync: statSync3 } = await import("fs");
6376
- const sizeBytes = statSync3(dest).size;
6375
+ const { statSync: statSync2 } = await import("fs");
6376
+ const sizeBytes = statSync2(dest).size;
6377
6377
  return { savedPath: dest, sizeBytes, mimeType };
6378
6378
  }
6379
6379
  async function harvestPageMedia(html, pageUrl, options = {}) {
@@ -7992,8 +7992,8 @@ var init_rates = __esm({
7992
7992
  "use strict";
7993
7993
  MC_COSTS = {
7994
7994
  serp: 100,
7995
- paa: 200,
7996
- paa_base: 1e3,
7995
+ paa: 300,
7996
+ paa_base: 5500,
7997
7997
  page_scrape: 100,
7998
7998
  url_map: 500,
7999
7999
  yt_channel: 200,
@@ -10648,8 +10648,8 @@ var init_blob_store = __esm({
10648
10648
  async get(key) {
10649
10649
  const path6 = (0, import_node_path2.join)(this.baseDir, "blobs", key);
10650
10650
  try {
10651
- const { readFileSync: readFileSync5 } = await import("fs");
10652
- return readFileSync5(path6);
10651
+ const { readFileSync: readFileSync6 } = await import("fs");
10652
+ return readFileSync6(path6);
10653
10653
  } catch {
10654
10654
  return null;
10655
10655
  }
@@ -15788,6 +15788,12 @@ var init_facebook_ad_routes = __esm({
15788
15788
  }
15789
15789
  }
15790
15790
  const results = [...advertiserMap.values()].sort((a, b) => b.adCount - a.adCount).slice(0, maxResults).map((a) => ({ name: a.pageName, pageName: a.pageName, libraryId: a.sampleLibraryId, sampleLibraryId: a.sampleLibraryId, adCount: a.adCount }));
15791
+ if (results.length === 0) {
15792
+ await creditMc(fbUser.id, MC_COSTS.fb_search, LedgerOperation.FB_SEARCH_REFUND, "empty result");
15793
+ searchRefunded = true;
15794
+ await logRequestEvent({ userId: fbUser.id, source: "facebook_search", status: "failed", query: body.query.trim(), error: "empty result refunded" });
15795
+ return c.json({ error: "no advertisers found (refunded)" }, 503);
15796
+ }
15791
15797
  const searchResult = { query: body.query.trim(), searchUrl, results };
15792
15798
  await logRequestEvent({ userId: fbUser.id, source: "facebook_search", status: "done", query: body.query.trim(), resultCount: results.length, result: searchResult });
15793
15799
  return c.json(searchResult);
@@ -16600,20 +16606,15 @@ async function resolveInstagramLaunch(body) {
16600
16606
  }
16601
16607
  };
16602
16608
  }
16603
- function outputBaseDir() {
16604
- return process.env.MCP_SCRAPER_OUTPUT_DIR?.trim() || (0, import_node_path7.join)((0, import_node_os5.homedir)(), "Downloads", "mcp-scraper");
16605
- }
16606
16609
  function safeFilePart(input) {
16607
16610
  return input.replace(/^https?:\/\//, "").replace(/[^a-z0-9._-]+/gi, "-").replace(/^-+|-+$/g, "").slice(0, 80) || "instagram";
16608
16611
  }
16609
- function mediaOutputDir(shortcode, sourceUrl) {
16612
+ function blobKeyPrefix(shortcode, sourceUrl) {
16610
16613
  const stamp = (/* @__PURE__ */ new Date()).toISOString().replace(/[:.]/g, "-").slice(0, 19);
16611
16614
  const slug = shortcode ? `instagram-${shortcode}` : safeFilePart(sourceUrl);
16612
- const dir = (0, import_node_path7.join)(outputBaseDir(), "instagram", `${stamp}-${slug}`);
16613
- (0, import_node_fs5.mkdirSync)(dir, { recursive: true });
16614
- return dir;
16615
+ return `instagram/${stamp}-${slug}`;
16615
16616
  }
16616
- async function downloadToFile(url, destPath, referer) {
16617
+ async function fetchMediaBytes(url, referer) {
16617
16618
  const res = await fetch(url, {
16618
16619
  headers: {
16619
16620
  "User-Agent": "Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/124.0.0.0 Safari/537.36",
@@ -16623,12 +16624,8 @@ async function downloadToFile(url, destPath, referer) {
16623
16624
  });
16624
16625
  if (!res.ok) throw new Error(`HTTP ${res.status}`);
16625
16626
  if (!res.body) throw new Error("Empty response body");
16626
- await (0, import_promises4.pipeline)(import_node_stream2.Readable.fromWeb(res.body), (0, import_node_fs5.createWriteStream)(destPath));
16627
- return {
16628
- savedPath: destPath,
16629
- sizeBytes: (0, import_node_fs5.statSync)(destPath).size,
16630
- mimeType: res.headers.get("content-type")?.split(";")[0]?.trim() ?? null
16631
- };
16627
+ const bytes = Buffer.from(await res.arrayBuffer());
16628
+ return { bytes, mimeType: res.headers.get("content-type")?.split(";")[0]?.trim() ?? null };
16632
16629
  }
16633
16630
  function extFromMime(mimeType, fallback) {
16634
16631
  const map = {
@@ -16661,7 +16658,7 @@ function trackFilename(track, index, shortcode) {
16661
16658
  const bitrate = track.bitrate ? `-${track.bitrate}` : "";
16662
16659
  return `${label}-${type}-${index}${bitrate}.mp4`;
16663
16660
  }
16664
- var import_hono7, import_zod18, import_node_fs5, import_node_os5, import_node_path7, import_node_stream2, import_promises4, import_node_child_process2, InstagramProfileContentBodySchema, InstagramMediaTypeSchema, InstagramMediaDownloadBodySchema, instagramApp;
16661
+ var import_hono7, import_zod18, import_node_fs5, import_node_os5, import_node_path7, import_node_child_process2, InstagramProfileContentBodySchema, InstagramMediaTypeSchema, InstagramMediaDownloadBodySchema, instagramApp;
16665
16662
  var init_instagram_routes = __esm({
16666
16663
  "src/api/instagram-routes.ts"() {
16667
16664
  "use strict";
@@ -16670,8 +16667,6 @@ var init_instagram_routes = __esm({
16670
16667
  import_node_fs5 = require("fs");
16671
16668
  import_node_os5 = require("os");
16672
16669
  import_node_path7 = require("path");
16673
- import_node_stream2 = require("stream");
16674
- import_promises4 = require("stream/promises");
16675
16670
  import_node_child_process2 = require("child_process");
16676
16671
  init_browser_service_env();
16677
16672
  init_BrowserDriver();
@@ -16680,6 +16675,7 @@ var init_instagram_routes = __esm({
16680
16675
  init_db();
16681
16676
  init_rates();
16682
16677
  init_concurrency_gates();
16678
+ init_blob_store();
16683
16679
  init_InstagramContentExtractor();
16684
16680
  init_media_transcription();
16685
16681
  InstagramProfileContentBodySchema = import_zod18.z.object({
@@ -16794,55 +16790,65 @@ var init_instagram_routes = __esm({
16794
16790
  const mediaTypes = new Set(body.mediaTypes);
16795
16791
  const downloads = [];
16796
16792
  const warnings = [];
16797
- let outputDir = null;
16798
16793
  if (body.downloadMedia) {
16799
- outputDir = mediaOutputDir(extraction.shortcode, sourceUrl.href);
16800
- const textPath = (0, import_node_path7.join)(outputDir, `${extraction.shortcode ?? "instagram"}-text.txt`);
16801
- (0, import_node_fs5.writeFileSync)(textPath, [
16802
- `URL: ${extraction.pageUrl}`,
16803
- extraction.caption ? `Caption: ${extraction.caption}` : "",
16804
- "",
16805
- extraction.bodyText
16806
- ].filter(Boolean).join("\n"), "utf8");
16807
- downloads.push({ kind: "text", url: null, savedPath: textPath, sizeBytes: (0, import_node_fs5.statSync)(textPath).size, mimeType: "text/plain", error: null });
16808
- if (mediaTypes.has("image") && extraction.imageUrl) {
16809
- const imagePathBase = (0, import_node_path7.join)(outputDir, `${extraction.shortcode ?? "instagram"}-image`);
16810
- try {
16811
- const downloaded = await downloadToFile(extraction.imageUrl, imagePathBase, extraction.pageUrl);
16812
- const ext = extFromMime(downloaded.mimeType, "jpg");
16813
- const finalPath = downloaded.savedPath.endsWith(`.${ext}`) ? downloaded.savedPath : `${downloaded.savedPath}.${ext}`;
16814
- if (finalPath !== downloaded.savedPath) {
16815
- const { renameSync } = await import("fs");
16816
- renameSync(downloaded.savedPath, finalPath);
16794
+ let muxWorkDir = null;
16795
+ try {
16796
+ const store = getBlobStore();
16797
+ const prefix = blobKeyPrefix(extraction.shortcode, sourceUrl.href);
16798
+ const textContent = [
16799
+ `URL: ${extraction.pageUrl}`,
16800
+ extraction.caption ? `Caption: ${extraction.caption}` : "",
16801
+ "",
16802
+ extraction.bodyText
16803
+ ].filter(Boolean).join("\n");
16804
+ const storedText = await store.put(`${prefix}/text.txt`, textContent, "text/plain");
16805
+ downloads.push({ kind: "text", url: null, savedPath: storedText.url, sizeBytes: storedText.bytes, mimeType: "text/plain", error: null });
16806
+ if (mediaTypes.has("image") && extraction.imageUrl) {
16807
+ try {
16808
+ const { bytes, mimeType } = await fetchMediaBytes(extraction.imageUrl, extraction.pageUrl);
16809
+ const ext = extFromMime(mimeType, "jpg");
16810
+ const stored = await store.put(`${prefix}/image.${ext}`, bytes, mimeType ?? "application/octet-stream");
16811
+ downloads.push({ kind: "image", url: extraction.imageUrl, savedPath: stored.url, sizeBytes: stored.bytes, mimeType, error: null });
16812
+ } catch (err) {
16813
+ downloads.push({ kind: "image", url: extraction.imageUrl, savedPath: null, sizeBytes: null, mimeType: null, error: err instanceof Error ? err.message : String(err) });
16817
16814
  }
16818
- downloads.push({ kind: "image", url: extraction.imageUrl, savedPath: finalPath, sizeBytes: (0, import_node_fs5.statSync)(finalPath).size, mimeType: downloaded.mimeType, error: null });
16819
- } catch (err) {
16820
- downloads.push({ kind: "image", url: extraction.imageUrl, savedPath: null, sizeBytes: null, mimeType: null, error: err instanceof Error ? err.message : String(err) });
16821
16815
  }
16822
- }
16823
- const tracksToDownload = body.downloadAllTracks ? extraction.tracks : [extraction.selectedVideoTrack, extraction.selectedAudioTrack].filter((track) => Boolean(track));
16824
- const savedByUrl = /* @__PURE__ */ new Map();
16825
- for (const [index, track] of tracksToDownload.entries()) {
16826
- if (track.streamType === "video" && !mediaTypes.has("video")) continue;
16827
- if (track.streamType === "audio" && !mediaTypes.has("audio")) continue;
16828
- const kind = track.streamType === "audio" ? "audio" : "video";
16829
- const target = (0, import_node_path7.join)(outputDir, trackFilename(track, index, extraction.shortcode));
16830
- try {
16831
- const downloaded = await downloadToFile(track.url, target, extraction.pageUrl);
16832
- downloads.push({ kind, url: track.url, savedPath: downloaded.savedPath, sizeBytes: downloaded.sizeBytes, mimeType: downloaded.mimeType, error: null });
16833
- savedByUrl.set(track.url, downloaded.savedPath);
16834
- } catch (err) {
16835
- downloads.push({ kind, url: track.url, savedPath: null, sizeBytes: null, mimeType: null, error: err instanceof Error ? err.message : String(err) });
16816
+ const tracksToDownload = body.downloadAllTracks ? extraction.tracks : [extraction.selectedVideoTrack, extraction.selectedAudioTrack].filter((track) => Boolean(track));
16817
+ const muxEligible = Boolean(body.mux && mediaTypes.has("video") && mediaTypes.has("audio") && extraction.selectedVideoTrack && extraction.selectedAudioTrack);
16818
+ const muxInputs = /* @__PURE__ */ new Map();
16819
+ for (const [index, track] of tracksToDownload.entries()) {
16820
+ if (track.streamType === "video" && !mediaTypes.has("video")) continue;
16821
+ if (track.streamType === "audio" && !mediaTypes.has("audio")) continue;
16822
+ const kind = track.streamType === "audio" ? "audio" : "video";
16823
+ try {
16824
+ const { bytes, mimeType } = await fetchMediaBytes(track.url, extraction.pageUrl);
16825
+ const stored = await store.put(`${prefix}/${trackFilename(track, index, extraction.shortcode)}`, bytes, mimeType ?? "video/mp4");
16826
+ downloads.push({ kind, url: track.url, savedPath: stored.url, sizeBytes: stored.bytes, mimeType, error: null });
16827
+ const isMuxInput = muxEligible && (track.url === extraction.selectedVideoTrack.url || track.url === extraction.selectedAudioTrack.url);
16828
+ if (isMuxInput) {
16829
+ if (!muxWorkDir) muxWorkDir = (0, import_node_fs5.mkdtempSync)((0, import_node_path7.join)((0, import_node_os5.tmpdir)(), "ig-mux-"));
16830
+ const localPath = (0, import_node_path7.join)(muxWorkDir, trackFilename(track, index, extraction.shortcode));
16831
+ (0, import_node_fs5.writeFileSync)(localPath, bytes);
16832
+ muxInputs.set(track.url, localPath);
16833
+ }
16834
+ } catch (err) {
16835
+ downloads.push({ kind, url: track.url, savedPath: null, sizeBytes: null, mimeType: null, error: err instanceof Error ? err.message : String(err) });
16836
+ }
16836
16837
  }
16837
- }
16838
- if (body.mux && mediaTypes.has("video") && mediaTypes.has("audio") && extraction.selectedVideoTrack && extraction.selectedAudioTrack && savedByUrl.has(extraction.selectedVideoTrack.url) && savedByUrl.has(extraction.selectedAudioTrack.url)) {
16839
- const outPath = (0, import_node_path7.join)(outputDir, `${extraction.shortcode ?? "instagram"}-muxed.mp4`);
16840
- const muxed = await runFfmpegMux(savedByUrl.get(extraction.selectedVideoTrack.url), savedByUrl.get(extraction.selectedAudioTrack.url), outPath);
16841
- if (muxed.ok) {
16842
- downloads.push({ kind: "muxed_video", url: null, savedPath: outPath, sizeBytes: (0, import_node_fs5.statSync)(outPath).size, mimeType: "video/mp4", error: null });
16843
- } else {
16844
- warnings.push(`Mux skipped/failed: ${muxed.error}`);
16838
+ if (muxEligible && muxWorkDir && muxInputs.has(extraction.selectedVideoTrack.url) && muxInputs.has(extraction.selectedAudioTrack.url)) {
16839
+ const outPath = (0, import_node_path7.join)(muxWorkDir, "muxed.mp4");
16840
+ const muxed = await runFfmpegMux(muxInputs.get(extraction.selectedVideoTrack.url), muxInputs.get(extraction.selectedAudioTrack.url), outPath);
16841
+ if (muxed.ok) {
16842
+ const stored = await store.put(`${prefix}/muxed.mp4`, (0, import_node_fs5.readFileSync)(outPath), "video/mp4");
16843
+ downloads.push({ kind: "muxed_video", url: null, savedPath: stored.url, sizeBytes: stored.bytes, mimeType: "video/mp4", error: null });
16844
+ } else {
16845
+ warnings.push(`Mux skipped/failed: ${muxed.error}`);
16846
+ }
16845
16847
  }
16848
+ } catch (err) {
16849
+ warnings.push(`Media download unavailable: ${err instanceof Error ? err.message : String(err)}`);
16850
+ } finally {
16851
+ if (muxWorkDir) (0, import_node_fs5.rmSync)(muxWorkDir, { recursive: true, force: true });
16846
16852
  }
16847
16853
  }
16848
16854
  let transcript = null;
@@ -16865,7 +16871,7 @@ var init_instagram_routes = __esm({
16865
16871
  const result = {
16866
16872
  ...extraction,
16867
16873
  browser: launch.browser,
16868
- outputDir,
16874
+ outputDir: null,
16869
16875
  downloads,
16870
16876
  warnings,
16871
16877
  transcript
@@ -17147,9 +17153,6 @@ async function memoryCall(toolName, args, userMemoryKey) {
17147
17153
  return { ok: false, error: err?.message ?? "memory call failed" };
17148
17154
  }
17149
17155
  }
17150
- function personalVault(user) {
17151
- return `mcp-${user.id}`;
17152
- }
17153
17156
  function memoryIdentity(user) {
17154
17157
  return user.email;
17155
17158
  }
@@ -19067,12 +19070,12 @@ function reportTitle(full) {
19067
19070
  const title = full.split("\n").find((line) => line.startsWith("# "));
19068
19071
  return title?.replace(/^#\s+/, "").trim() || "MCP Scraper Report";
19069
19072
  }
19070
- function outputBaseDir2() {
19073
+ function outputBaseDir() {
19071
19074
  return process.env.MCP_SCRAPER_OUTPUT_DIR?.trim() || (0, import_node_path8.join)((0, import_node_os6.homedir)(), "Downloads", "mcp-scraper");
19072
19075
  }
19073
19076
  function saveFullReport(full) {
19074
19077
  if (!reportSavingEnabled || process.env.MCP_SCRAPER_SAVE_REPORTS === "false") return null;
19075
- const outDir = outputBaseDir2();
19078
+ const outDir = outputBaseDir();
19076
19079
  try {
19077
19080
  (0, import_node_fs6.mkdirSync)(outDir, { recursive: true });
19078
19081
  const stamp = (/* @__PURE__ */ new Date()).toISOString().replace(/[:.]/g, "-");
@@ -19090,7 +19093,7 @@ function saveBulkSite(siteUrl, pages, seo, imageAudit) {
19090
19093
  if (!reportSavingActive()) return null;
19091
19094
  try {
19092
19095
  const stamp = (/* @__PURE__ */ new Date()).toISOString().replace(/[:.]/g, "-");
19093
- const dir = (0, import_node_path8.join)(outputBaseDir2(), `extract-${slugifyReportName(siteUrl)}-${stamp}`);
19096
+ const dir = (0, import_node_path8.join)(outputBaseDir(), `extract-${slugifyReportName(siteUrl)}-${stamp}`);
19094
19097
  const pagesDir = (0, import_node_path8.join)(dir, "pages");
19095
19098
  (0, import_node_fs6.mkdirSync)(pagesDir, { recursive: true });
19096
19099
  const indexRows = pages.map((p, i) => {
@@ -19183,7 +19186,7 @@ ${renderImageSection(imageAudit)}` : ""), "utf8");
19183
19186
  function saveUrlInventory(siteUrl, urls) {
19184
19187
  if (!reportSavingActive()) return null;
19185
19188
  try {
19186
- const outDir = outputBaseDir2();
19189
+ const outDir = outputBaseDir();
19187
19190
  (0, import_node_fs6.mkdirSync)(outDir, { recursive: true });
19188
19191
  const stamp = (/* @__PURE__ */ new Date()).toISOString().replace(/[:.]/g, "-");
19189
19192
  const file = (0, import_node_path8.join)(outDir, `${stamp}-urlmap-${slugifyReportName(siteUrl.replace(/^https?:\/\//, ""))}.csv`);
@@ -19198,7 +19201,7 @@ function saveUrlInventory(siteUrl, urls) {
19198
19201
  function persistScreenshotLocally(base64, url) {
19199
19202
  if (!reportSavingEnabled || process.env.MCP_SCRAPER_SAVE_REPORTS === "false") return null;
19200
19203
  try {
19201
- const dir = (0, import_node_path8.join)(outputBaseDir2(), "screenshots");
19204
+ const dir = (0, import_node_path8.join)(outputBaseDir(), "screenshots");
19202
19205
  (0, import_node_fs6.mkdirSync)(dir, { recursive: true });
19203
19206
  const stamp = (/* @__PURE__ */ new Date()).toISOString().replace(/[:.]/g, "-");
19204
19207
  const slug = url.replace(/^https?:\/\//, "").replace(/[^a-z0-9]+/gi, "-").replace(/^-+|-+$/g, "").slice(0, 60);
@@ -21478,7 +21481,7 @@ function localLocationFileAllowed() {
21478
21481
  async function existingPath(value) {
21479
21482
  const trimmed = value?.trim();
21480
21483
  if (!trimmed) return null;
21481
- await (0, import_promises5.access)(trimmed);
21484
+ await (0, import_promises4.access)(trimmed);
21482
21485
  return trimmed;
21483
21486
  }
21484
21487
  async function resolveUsZipsPath(requestedPath) {
@@ -21496,7 +21499,7 @@ async function loadZipGroups(stateAbbr, requestedPath, warnings) {
21496
21499
  }
21497
21500
  const path6 = await resolveUsZipsPath(requestedPath);
21498
21501
  if (!path6) return { path: null, groups: /* @__PURE__ */ new Map() };
21499
- const records = csvRecords(await (0, import_promises5.readFile)(path6, "utf8"));
21502
+ const records = csvRecords(await (0, import_promises4.readFile)(path6, "utf8"));
21500
21503
  const groups = /* @__PURE__ */ new Map();
21501
21504
  for (const record of records) {
21502
21505
  const state = (record.state_abbr ?? record.state ?? "").trim().toUpperCase();
@@ -21547,11 +21550,11 @@ async function resolveDirectoryMarkets(options) {
21547
21550
  }
21548
21551
  return { markets, censusSourceUrl: sourceUrl, usZipsSourcePath: zipData.path, warnings };
21549
21552
  }
21550
- var import_promises5, POPULATION_YEARS, STATE_META, STATE_BY_NAME;
21553
+ var import_promises4, POPULATION_YEARS, STATE_META, STATE_BY_NAME;
21551
21554
  var init_location_db = __esm({
21552
21555
  "src/directory/location-db.ts"() {
21553
21556
  "use strict";
21554
- import_promises5 = require("fs/promises");
21557
+ import_promises4 = require("fs/promises");
21555
21558
  init_csv();
21556
21559
  POPULATION_YEARS = [2020, 2021, 2022, 2023, 2024, 2025];
21557
21560
  STATE_META = {
@@ -21767,8 +21770,8 @@ function csvRowsFor(result) {
21767
21770
  return rows;
21768
21771
  }
21769
21772
  async function saveDirectoryCsv(result) {
21770
- const outDir = (0, import_node_path9.join)(outputBaseDir2(), "directory-workflows");
21771
- await (0, import_promises6.mkdir)(outDir, { recursive: true });
21773
+ const outDir = (0, import_node_path9.join)(outputBaseDir(), "directory-workflows");
21774
+ await (0, import_promises5.mkdir)(outDir, { recursive: true });
21772
21775
  const stamp = result.extractedAt.replace(/[:.]/g, "-");
21773
21776
  const slug = `${result.state}-${result.query}`.toLowerCase().replace(/[^a-z0-9]+/g, "-").replace(/^-+|-+$/g, "").slice(0, 80);
21774
21777
  const path6 = (0, import_node_path9.join)(outDir, `${stamp}-${slug}-directory-workflow.csv`);
@@ -21803,7 +21806,7 @@ async function saveDirectoryCsv(result) {
21803
21806
  "extracted_at",
21804
21807
  "duration_ms"
21805
21808
  ];
21806
- await (0, import_promises6.writeFile)(path6, rowsToCsv(headers, csvRowsFor(result)), "utf8");
21809
+ await (0, import_promises5.writeFile)(path6, rowsToCsv(headers, csvRowsFor(result)), "utf8");
21807
21810
  return path6;
21808
21811
  }
21809
21812
  async function runDirectoryWorkflowFromPlan(options, plan) {
@@ -21829,11 +21832,11 @@ async function runDirectoryWorkflowFromPlan(options, plan) {
21829
21832
  const csvPath = options.saveCsv ? await saveDirectoryCsv(base) : null;
21830
21833
  return { ...base, csvPath };
21831
21834
  }
21832
- var import_promises6, import_node_path9, import_zod24, DirectoryWorkflowOptionsSchema;
21835
+ var import_promises5, import_node_path9, import_zod24, DirectoryWorkflowOptionsSchema;
21833
21836
  var init_directory_workflow = __esm({
21834
21837
  "src/directory/directory-workflow.ts"() {
21835
21838
  "use strict";
21836
- import_promises6 = require("fs/promises");
21839
+ import_promises5 = require("fs/promises");
21837
21840
  import_node_path9 = require("path");
21838
21841
  import_zod24 = require("zod");
21839
21842
  init_mcp_response_formatter();
@@ -22001,7 +22004,7 @@ async function readIndex(baseDir) {
22001
22004
  const path6 = indexPath(baseDir);
22002
22005
  if (!(0, import_node_fs7.existsSync)(path6)) return { runs: [] };
22003
22006
  try {
22004
- return JSON.parse(await (0, import_promises7.readFile)(path6, "utf8"));
22007
+ return JSON.parse(await (0, import_promises6.readFile)(path6, "utf8"));
22005
22008
  } catch {
22006
22009
  return { runs: [] };
22007
22010
  }
@@ -22021,14 +22024,14 @@ async function updateWorkflowIndex(manifest, manifestPath, baseDir) {
22021
22024
  summary: `${manifest.title} (${manifest.status})`
22022
22025
  };
22023
22026
  index.runs = [entry, ...index.runs.filter((r) => r.runId !== manifest.runId)].slice(0, 200);
22024
- await (0, import_promises7.mkdir)((0, import_node_path10.dirname)(path6), { recursive: true });
22025
- await (0, import_promises7.writeFile)(path6, JSON.stringify(index, null, 2), "utf8");
22027
+ await (0, import_promises6.mkdir)((0, import_node_path10.dirname)(path6), { recursive: true });
22028
+ await (0, import_promises6.writeFile)(path6, JSON.stringify(index, null, 2), "utf8");
22026
22029
  }
22027
- var import_promises7, import_node_fs7, import_node_os7, import_node_path10, import_node_child_process3, ArtifactWriter;
22030
+ var import_promises6, import_node_fs7, import_node_os7, import_node_path10, import_node_child_process3, ArtifactWriter;
22028
22031
  var init_artifact_writer = __esm({
22029
22032
  "src/workflows/artifact-writer.ts"() {
22030
22033
  "use strict";
22031
- import_promises7 = require("fs/promises");
22034
+ import_promises6 = require("fs/promises");
22032
22035
  import_node_fs7 = require("fs");
22033
22036
  import_node_os7 = require("os");
22034
22037
  import_node_path10 = require("path");
@@ -22060,31 +22063,31 @@ var init_artifact_writer = __esm({
22060
22063
  const nameSource = String(input.keyword ?? input.query ?? input.domain ?? input.state ?? workflowId);
22061
22064
  const baseDir = workflowOutputBaseDir(outputDir);
22062
22065
  const runDir = (0, import_node_path10.join)(baseDir, "workflows", workflowId, `${timestamp()}-${safeSlug(nameSource)}-${safeSlug(runId).slice(0, 20)}`);
22063
- await (0, import_promises7.mkdir)(runDir, { recursive: true });
22066
+ await (0, import_promises6.mkdir)(runDir, { recursive: true });
22064
22067
  return new _ArtifactWriter(workflowId, title, runId, baseDir, runDir, startedAt, input);
22065
22068
  }
22066
22069
  async remember(kind, label, path6, rows) {
22067
- const size = await (0, import_promises7.stat)(path6).then((s) => s.size).catch(() => void 0);
22070
+ const size = await (0, import_promises6.stat)(path6).then((s) => s.size).catch(() => void 0);
22068
22071
  this.artifacts.push({ kind, label, path: path6, bytes: size, rows });
22069
22072
  await postToMemoryLibrary({ title: `${this.title} ${label}`, content: `artifact: ${path6}`, source: `mcp-scraper-workflow:${this.workflowId}:${this.runId}` });
22070
22073
  return path6;
22071
22074
  }
22072
22075
  async writeJson(label, relativePath, data) {
22073
22076
  const path6 = (0, import_node_path10.join)(this.runDir, relativePath);
22074
- await (0, import_promises7.mkdir)((0, import_node_path10.dirname)(path6), { recursive: true });
22075
- await (0, import_promises7.writeFile)(path6, JSON.stringify(data, null, 2), "utf8");
22077
+ await (0, import_promises6.mkdir)((0, import_node_path10.dirname)(path6), { recursive: true });
22078
+ await (0, import_promises6.writeFile)(path6, JSON.stringify(data, null, 2), "utf8");
22076
22079
  return this.remember("json", label, path6);
22077
22080
  }
22078
22081
  async writeText(label, relativePath, text, kind = "markdown") {
22079
22082
  const path6 = (0, import_node_path10.join)(this.runDir, relativePath);
22080
- await (0, import_promises7.mkdir)((0, import_node_path10.dirname)(path6), { recursive: true });
22081
- await (0, import_promises7.writeFile)(path6, text, "utf8");
22083
+ await (0, import_promises6.mkdir)((0, import_node_path10.dirname)(path6), { recursive: true });
22084
+ await (0, import_promises6.writeFile)(path6, text, "utf8");
22082
22085
  return this.remember(kind, label, path6);
22083
22086
  }
22084
22087
  async writeCsv(label, relativePath, headers, rows) {
22085
22088
  const path6 = (0, import_node_path10.join)(this.runDir, relativePath);
22086
- await (0, import_promises7.mkdir)((0, import_node_path10.dirname)(path6), { recursive: true });
22087
- await (0, import_promises7.writeFile)(path6, rowsToCsv(headers, rows), "utf8");
22089
+ await (0, import_promises6.mkdir)((0, import_node_path10.dirname)(path6), { recursive: true });
22090
+ await (0, import_promises6.writeFile)(path6, rowsToCsv(headers, rows), "utf8");
22088
22091
  return this.remember("csv", label, path6, rows.length);
22089
22092
  }
22090
22093
  async writeHtml(label, relativePath, html) {
@@ -22105,7 +22108,7 @@ var init_artifact_writer = __esm({
22105
22108
  counts
22106
22109
  };
22107
22110
  const path6 = (0, import_node_path10.join)(this.runDir, "manifest.json");
22108
- await (0, import_promises7.writeFile)(path6, JSON.stringify(manifest, null, 2), "utf8");
22111
+ await (0, import_promises6.writeFile)(path6, JSON.stringify(manifest, null, 2), "utf8");
22109
22112
  await updateWorkflowIndex(manifest, path6, this.baseDir);
22110
22113
  return path6;
22111
22114
  }
@@ -23831,7 +23834,7 @@ async function runWorkflowStep(id, rawInput, opts) {
23831
23834
  const writtenArtifacts = await Promise.all(
23832
23835
  artifacts.artifacts.map(async (artifact) => ({
23833
23836
  ...artifact,
23834
- content: await (0, import_promises8.readFile)(artifact.path, "utf8").catch(() => "")
23837
+ content: await (0, import_promises7.readFile)(artifact.path, "utf8").catch(() => "")
23835
23838
  }))
23836
23839
  );
23837
23840
  return {
@@ -23850,11 +23853,11 @@ async function runWorkflowStep(id, rawInput, opts) {
23850
23853
  artifacts: writtenArtifacts
23851
23854
  };
23852
23855
  }
23853
- var import_promises8, import_zod30, DEFINITIONS;
23856
+ var import_promises7, import_zod30, DEFINITIONS;
23854
23857
  var init_registry2 = __esm({
23855
23858
  "src/workflows/registry.ts"() {
23856
23859
  "use strict";
23857
- import_promises8 = require("fs/promises");
23860
+ import_promises7 = require("fs/promises");
23858
23861
  import_zod30 = require("zod");
23859
23862
  init_artifact_writer();
23860
23863
  init_http_client2();
@@ -23925,7 +23928,7 @@ async function readManifestFromSummary(summary) {
23925
23928
  if (!summary.reportPath) return null;
23926
23929
  const manifestPath = summary.reportPath.replace(/report\.html$/, "manifest.json");
23927
23930
  try {
23928
- return JSON.parse(await (0, import_promises9.readFile)(manifestPath, "utf8"));
23931
+ return JSON.parse(await (0, import_promises8.readFile)(manifestPath, "utf8"));
23929
23932
  } catch {
23930
23933
  return null;
23931
23934
  }
@@ -24140,12 +24143,12 @@ async function dispatchDueWorkflowSchedules(apiUrl, limit = 3) {
24140
24143
  }
24141
24144
  return { dispatched: results.length, results };
24142
24145
  }
24143
- var import_node_crypto8, import_promises9, import_hono14, import_zod31, workflowApp, WorkflowInputSchema, WorkflowIdSchema, CadenceSchema, ScheduleStatusSchema, RunBodySchema, ScheduleCreateSchema, SchedulePatchSchema, TERMINAL_RUN_STATUSES;
24146
+ var import_node_crypto8, import_promises8, import_hono14, import_zod31, workflowApp, WorkflowInputSchema, WorkflowIdSchema, CadenceSchema, ScheduleStatusSchema, RunBodySchema, ScheduleCreateSchema, SchedulePatchSchema, TERMINAL_RUN_STATUSES;
24144
24147
  var init_workflow_routes = __esm({
24145
24148
  "src/api/workflow-routes.ts"() {
24146
24149
  "use strict";
24147
24150
  import_node_crypto8 = require("crypto");
24148
- import_promises9 = require("fs/promises");
24151
+ import_promises8 = require("fs/promises");
24149
24152
  import_hono14 = require("hono");
24150
24153
  import_zod31 = require("zod");
24151
24154
  init_artifact_writer();
@@ -24315,7 +24318,7 @@ var init_workflow_routes = __esm({
24315
24318
  return new Response(artifact.content, { headers });
24316
24319
  }
24317
24320
  try {
24318
- const content = await (0, import_promises9.readFile)(artifact.path);
24321
+ const content = await (0, import_promises8.readFile)(artifact.path);
24319
24322
  return new Response(content, { headers });
24320
24323
  } catch {
24321
24324
  return c.json({ error: "Artifact file is no longer available" }, 410);
@@ -25254,13 +25257,41 @@ var init_PAAExtractor = __esm({
25254
25257
  }
25255
25258
  return parts.join(" ").replace(/\s+/g, " ").trim();
25256
25259
  }
25257
- return Array.from(document.querySelectorAll(selectors.item)).map((pair) => ({
25258
- question: pair.getAttribute(selectors.itemDataQ) || pair.getAttribute(selectors.itemDataInitQ) || pair.querySelector(selectors.itemQuestionEl)?.innerText?.trim() || "",
25259
- answer: cleanText(pair.querySelector(selectors.answerContainer)) || void 0,
25260
- sourceTitle: pair.querySelector(selectors.sourceTitle)?.innerText?.trim() || void 0,
25261
- sourceSite: pair.querySelector(selectors.sourceSite)?.innerText?.trim() || void 0,
25262
- sourceCite: pair.querySelector(selectors.sourceCite)?.innerText?.trim() || void 0
25263
- }));
25260
+ return Array.from(document.querySelectorAll(selectors.item)).map((pair) => {
25261
+ let sourceTitle = pair.querySelector(selectors.sourceTitle)?.innerText?.trim() || void 0;
25262
+ let sourceSite = pair.querySelector(selectors.sourceSite)?.innerText?.trim() || void 0;
25263
+ let sourceCite = pair.querySelector(selectors.sourceCite)?.innerText?.trim() || void 0;
25264
+ if (!sourceCite) {
25265
+ const seenHrefs = /* @__PURE__ */ new Set();
25266
+ const anchors = [];
25267
+ for (const a of Array.from(pair.querySelectorAll("a[href]"))) {
25268
+ const href = a.href;
25269
+ if (!/^https?:\/\//.test(href) || seenHrefs.has(href)) continue;
25270
+ let host = "";
25271
+ try {
25272
+ host = new URL(href).hostname;
25273
+ } catch {
25274
+ continue;
25275
+ }
25276
+ if (/(^|\.)google(usercontent)?\.[a-z.]+$/.test(host)) continue;
25277
+ seenHrefs.add(href);
25278
+ anchors.push({ text: (a.textContent || "").trim(), href, host });
25279
+ }
25280
+ const primary = anchors.find((x) => x.text.length > 2) ?? anchors[0];
25281
+ if (primary) {
25282
+ sourceTitle = sourceTitle || primary.text || void 0;
25283
+ sourceSite = sourceSite || primary.host.replace(/^www\./, "");
25284
+ sourceCite = primary.href;
25285
+ }
25286
+ }
25287
+ return {
25288
+ question: pair.getAttribute(selectors.itemDataQ) || pair.getAttribute(selectors.itemDataInitQ) || pair.querySelector(selectors.itemQuestionEl)?.innerText?.trim() || "",
25289
+ answer: cleanText(pair.querySelector(selectors.answerContainer)) || void 0,
25290
+ sourceTitle,
25291
+ sourceSite,
25292
+ sourceCite
25293
+ };
25294
+ });
25264
25295
  }, sels);
25265
25296
  return raw.flatMap((item) => {
25266
25297
  const cleaned = {
@@ -25302,54 +25333,115 @@ var init_PAAExtractor = __esm({
25302
25333
  const seenQs = /* @__PURE__ */ new Set();
25303
25334
  const orderedQs = [];
25304
25335
  const results = [];
25305
- const readAllQs = () => page.evaluate(
25306
- ({ sel, dataQ, dataInitQ, questionEl }) => Array.from(document.querySelectorAll(sel)).map(
25307
- (el2) => el2.getAttribute(dataQ) || el2.getAttribute(dataInitQ) || el2.querySelector(questionEl)?.innerText?.trim() || ""
25308
- ).filter(Boolean),
25309
- { sel: PAASelectors.item, dataQ: PAASelectors.itemDataQ, dataInitQ: PAASelectors.itemDataInitQ, questionEl: PAASelectors.itemQuestionEl }
25336
+ const abandonedNeverReclick = /* @__PURE__ */ new Set();
25337
+ const clickedOnceEver = /* @__PURE__ */ new Set();
25338
+ const itemArgs = {
25339
+ sel: PAASelectors.item,
25340
+ dataQ: PAASelectors.itemDataQ,
25341
+ dataInitQ: PAASelectors.itemDataInitQ,
25342
+ questionEl: PAASelectors.itemQuestionEl,
25343
+ answerSel: PAASelectors.answerContainer,
25344
+ expandedClass: PAASelectors.expandedClass,
25345
+ clickTarget: PAASelectors.clickTarget
25346
+ };
25347
+ const readItemStates = () => page.evaluate(
25348
+ ({ sel, dataQ, dataInitQ, questionEl, expandedClass, answerSel }) => {
25349
+ const answerText = (root) => {
25350
+ if (!root) return "";
25351
+ let out = "";
25352
+ const walk = (n) => {
25353
+ for (const c of Array.from(n.childNodes)) {
25354
+ if (c.nodeType === Node.TEXT_NODE) out += c.textContent || "";
25355
+ else if (c.nodeType === Node.ELEMENT_NODE && c.tagName !== "STYLE" && c.tagName !== "SCRIPT") walk(c);
25356
+ }
25357
+ };
25358
+ walk(root);
25359
+ return out.replace(/\s+/g, " ").trim();
25360
+ };
25361
+ return Array.from(document.querySelectorAll(sel)).map((el2) => ({
25362
+ q: el2.getAttribute(dataQ) || el2.getAttribute(dataInitQ) || el2.querySelector(questionEl)?.innerText?.trim() || "",
25363
+ done: el2.classList.contains(expandedClass) || answerText(el2.querySelector(answerSel)).length > 30
25364
+ })).filter((s) => s.q);
25365
+ },
25366
+ itemArgs
25310
25367
  );
25368
+ const expandOneItemSerially = async (q) => {
25369
+ const clicked = await page.evaluate(
25370
+ ({ sel, dataQ, dataInitQ, clickTarget, q: target }) => {
25371
+ const item = Array.from(document.querySelectorAll(sel)).find((el2) => (el2.getAttribute(dataQ) || el2.getAttribute(dataInitQ)) === target);
25372
+ const btn = item?.querySelector(clickTarget);
25373
+ if (!btn) return false;
25374
+ btn.scrollIntoView({ block: "center" });
25375
+ btn.click();
25376
+ return true;
25377
+ },
25378
+ { ...itemArgs, q }
25379
+ ).catch(() => false);
25380
+ if (!clicked) return "failed";
25381
+ try {
25382
+ const outcome = await page.waitForFunction(
25383
+ ({ sel, dataQ, dataInitQ, answerSel, q: target }) => {
25384
+ const answerText = (root) => {
25385
+ if (!root) return "";
25386
+ let out = "";
25387
+ const walk = (n) => {
25388
+ for (const c of Array.from(n.childNodes)) {
25389
+ if (c.nodeType === Node.TEXT_NODE) out += c.textContent || "";
25390
+ else if (c.nodeType === Node.ELEMENT_NODE && c.tagName !== "STYLE" && c.tagName !== "SCRIPT") walk(c);
25391
+ }
25392
+ };
25393
+ walk(root);
25394
+ return out.replace(/\s+/g, " ").trim();
25395
+ };
25396
+ const item = Array.from(document.querySelectorAll(sel)).find((el2) => (el2.getAttribute(dataQ) || el2.getAttribute(dataInitQ)) === target);
25397
+ if (!item) return "gone";
25398
+ const text = answerText(item.querySelector(answerSel));
25399
+ if (/an error has occurred/i.test(text.slice(0, 80))) return "error";
25400
+ if (text.length > 30) return "ok";
25401
+ return false;
25402
+ },
25403
+ { ...itemArgs, q },
25404
+ { timeout: 6500, polling: 200 }
25405
+ );
25406
+ return await outcome.jsonValue() === "ok" ? "ok" : "failed";
25407
+ } catch {
25408
+ return "failed";
25409
+ }
25410
+ };
25311
25411
  let round = 0;
25312
- while (seenQs.size < options.maxQuestions) {
25412
+ let growthWaits = 0;
25413
+ while (true) {
25313
25414
  if (options.softDeadlineMs && Date.now() >= options.softDeadlineMs) break;
25314
25415
  this.throwIfAborted(signal);
25315
- await this.throwIfCaptcha(page, "Google PAA expansion");
25316
- const beforeQs = await readAllQs();
25317
- if (beforeQs.length >= options.maxQuestions) break;
25318
- const unexpandedSel = `${PAASelectors.item}:not(.${PAASelectors.expandedClass}) ${PAASelectors.clickTarget}`;
25319
- const unexpandedCount = await page.locator(unexpandedSel).count();
25320
- if (unexpandedCount === 0) break;
25321
- this.reporter.onDepth(++round);
25322
- for (let ci = 0; ci < unexpandedCount; ci++) {
25323
- if (options.softDeadlineMs && Date.now() >= options.softDeadlineMs) break;
25324
- this.throwIfAborted(signal);
25325
- try {
25326
- const btn = page.locator(unexpandedSel).first();
25327
- await btn.scrollIntoViewIfNeeded();
25328
- await btn.hover({ force: true });
25329
- await page.waitForTimeout(100);
25330
- await btn.click({ force: true });
25331
- await page.waitForTimeout(500);
25332
- } catch {
25416
+ const states = await readItemStates();
25417
+ for (const s of states) {
25418
+ if (!seenQs.has(s.q)) {
25419
+ seenQs.add(s.q);
25420
+ orderedQs.push(s.q);
25333
25421
  }
25334
25422
  }
25335
- await page.waitForFunction(
25336
- ({ sel, min }) => document.querySelectorAll(sel).length > min,
25337
- { sel: PAASelectors.item, min: beforeQs.length },
25338
- { timeout: 5e3 }
25339
- ).catch(() => {
25340
- });
25341
- await this.throwIfCaptcha(page, "Google PAA expansion");
25342
- const afterQs = await readAllQs();
25343
- if (afterQs.length === beforeQs.length) break;
25344
- for (const q of afterQs) {
25345
- if (!seenQs.has(q)) {
25346
- seenQs.add(q);
25347
- orderedQs.push(q);
25348
- }
25423
+ const kept = new Set(orderedQs.slice(0, options.maxQuestions));
25424
+ const clickable = states.filter((s) => kept.has(s.q) && !clickedOnceEver.has(s.q) && !abandonedNeverReclick.has(s.q));
25425
+ const target = clickable.find((s) => !s.done) ?? (seenQs.size < options.maxQuestions ? clickable[0] : void 0);
25426
+ if (!target) {
25427
+ if (seenQs.size >= options.maxQuestions) break;
25428
+ if (++growthWaits > 1) break;
25429
+ const grew = await page.waitForFunction(
25430
+ ({ sel, min }) => document.querySelectorAll(sel).length > min,
25431
+ { sel: PAASelectors.item, min: states.length },
25432
+ { timeout: 3e3 }
25433
+ ).catch(() => null);
25434
+ if (!grew) break;
25435
+ continue;
25349
25436
  }
25437
+ growthWaits = 0;
25438
+ this.reporter.onDepth(++round);
25439
+ await this.throwIfCaptcha(page, "Google PAA expansion");
25440
+ clickedOnceEver.add(target.q);
25441
+ if (await expandOneItemSerially(target.q) === "failed") abandonedNeverReclick.add(target.q);
25350
25442
  }
25351
25443
  const itemMap = new Map((await this.extractVisibleItems(page)).map((i) => [i.question, i]));
25352
- await this.fillIncompleteItems(page, orderedQs, itemMap, options);
25444
+ await this.fillIncompleteItems(page, orderedQs, itemMap, options, abandonedNeverReclick);
25353
25445
  for (const q of orderedQs) {
25354
25446
  if (results.length >= options.maxQuestions) break;
25355
25447
  const key = this.normalizeQuestion(q);
@@ -25367,9 +25459,11 @@ var init_PAAExtractor = __esm({
25367
25459
  this.completeness.paaWithoutSource = results.filter((r) => !r.source_title && !r.source_site && !r.source_cite).length;
25368
25460
  return results;
25369
25461
  }
25370
- async fillIncompleteItems(page, orderedQs, itemMap, options) {
25462
+ async fillIncompleteItems(page, orderedQs, itemMap, options, abandonedNeverReclick) {
25371
25463
  const kept = new Set(orderedQs.slice(0, options.maxQuestions));
25372
- const missingBefore = new Set([...kept].filter((q) => !itemMap.get(q)?.answer));
25464
+ const missingBefore = new Set(
25465
+ [...kept].filter((q) => !itemMap.get(q)?.answer && !abandonedNeverReclick?.has(q))
25466
+ );
25373
25467
  if (missingBefore.size === 0) return;
25374
25468
  for (let round = 0; round < 2; round++) {
25375
25469
  if (options.softDeadlineMs && Date.now() >= options.softDeadlineMs) break;
@@ -25761,40 +25855,41 @@ var init_PAAExtractor = __esm({
25761
25855
  };
25762
25856
  }
25763
25857
  async captureAIOverviewShareUrl(page, options) {
25764
- for (let attempt = 0; attempt < 3; attempt++) {
25765
- if (options?.softDeadlineMs && Date.now() >= options.softDeadlineMs) return null;
25766
- const clicked = await page.evaluate(() => {
25767
- const heads = Array.from(document.querySelectorAll('h1,h2,h3,[role="heading"],div')).filter((h) => (h.textContent || "").trim() === "AI Overview");
25768
- for (const head of heads) {
25769
- let sec = head;
25770
- for (let up = 0; up < 8 && sec; up++) {
25771
- const btn = sec.querySelector('[aria-label="Share"]');
25772
- if (btn) {
25773
- btn.scrollIntoView({ block: "center" });
25774
- btn.click();
25775
- return true;
25776
- }
25777
- sec = sec.parentElement;
25858
+ if (options?.softDeadlineMs && Date.now() >= options.softDeadlineMs) return null;
25859
+ const clicked = await page.evaluate(() => {
25860
+ const heads = Array.from(document.querySelectorAll('h1,h2,h3,[role="heading"],div')).filter((h) => (h.textContent || "").trim() === "AI Overview");
25861
+ for (const head of heads) {
25862
+ let sec = head;
25863
+ for (let up = 0; up < 8 && sec; up++) {
25864
+ const btn = sec.querySelector('[aria-label="Share"]');
25865
+ if (btn) {
25866
+ btn.scrollIntoView({ block: "center" });
25867
+ btn.click();
25868
+ return true;
25778
25869
  }
25870
+ sec = sec.parentElement;
25779
25871
  }
25780
- return false;
25781
- }).catch(() => false);
25782
- if (!clicked) return null;
25783
- await page.waitForTimeout(4500);
25784
- const url = await page.evaluate(() => {
25872
+ }
25873
+ return false;
25874
+ }).catch(() => false);
25875
+ if (!clicked) return null;
25876
+ let url = null;
25877
+ try {
25878
+ const found = await page.waitForFunction(() => {
25785
25879
  const w = document.createTreeWalker(document.body, NodeFilter.SHOW_TEXT);
25786
25880
  while (w.nextNode()) {
25787
25881
  const m = (w.currentNode.textContent || "").match(/https:\/\/share\.google\/aimode\/\S+/);
25788
25882
  if (m) return m[0];
25789
25883
  }
25790
- return null;
25791
- }).catch(() => null);
25792
- await page.keyboard.press("Escape").catch(() => {
25793
- });
25794
- if (url) return url;
25795
- await page.waitForTimeout(600);
25884
+ return false;
25885
+ }, void 0, { timeout: 4500, polling: 250 });
25886
+ url = await found.jsonValue();
25887
+ } catch {
25888
+ url = null;
25796
25889
  }
25797
- return null;
25890
+ await page.keyboard.press("Escape").catch(() => {
25891
+ });
25892
+ return url;
25798
25893
  }
25799
25894
  buildTree(flat, _seed) {
25800
25895
  const roots = [];
@@ -27305,7 +27400,7 @@ var PACKAGE_VERSION;
27305
27400
  var init_version = __esm({
27306
27401
  "src/version.ts"() {
27307
27402
  "use strict";
27308
- PACKAGE_VERSION = "0.5.0";
27403
+ PACKAGE_VERSION = "0.7.0";
27309
27404
  }
27310
27405
  });
27311
27406
 
@@ -27404,7 +27499,7 @@ Multi-step orchestrations \u2014 prefer these over hand-chaining primitives when
27404
27499
  read it back for full detail rather than expecting the whole payload inline.
27405
27500
 
27406
27501
  ## Memory
27407
- mcp-scraper also exposes 74 persistent per-user memory tools (notes, facts, vaults,
27502
+ mcp-scraper also exposes persistent per-user memory tools (notes, facts, vaults,
27408
27503
  scheduled actions, tables, channels) backed by memory.mcpscraper.dev. Every account starts with 13
27409
27504
  vaults \u2014 call **list-vaults** to see what exists before creating anything new. Pick the vault whose job
27410
27505
  matches the content: **Ideas** (unvalidated concepts), **Knowledge** (distilled lessons/how-tos),
@@ -27414,11 +27509,45 @@ matches the content: **Ideas** (unvalidated concepts), **Knowledge** (distilled
27414
27509
  **Improvement Log** (receipts of changes made and whether they worked), **Experiments** (hypothesis +
27415
27510
  measured result), **Sprint** (current cycle's scope), **Inspiration** (other people's work worth
27416
27511
  studying).
27417
- Before writing, call **memory-suggest** to check whether something's already captured and to find the
27418
- right vault's hub note to link. Use **memory-put** for narrative notes (prose, a title, a body); use
27512
+ For a normal new narrative note, call **prepare-memory-write** first. It queries the live vault contract,
27513
+ tag vocabulary, and natural neighbor vaults, and returns interlink candidates that must be read before
27514
+ acceptance. Compose every returned template section, then call **memory-capture**, which validates the
27515
+ note, registers canonical tags, writes it, and verifies readback. Reserve low-level **memory-put** for
27516
+ deliberate edits and migrations. Use
27419
27517
  **table-create**/**table-insert-rows**/**table-query** instead when the caller wants row-shaped data
27420
27518
  they'll filter/sort by exact value. **memory-search** is meaning-based over full content (embeds, slower);
27421
27519
  **memory-list** filters one vault by kind/tags (fast, exact).
27520
+
27521
+ Treat every memory create or update as a graph operation. **prepare-memory-write** supplies the initial
27522
+ shortlist; use **memory-search** when it is weak or the topic is broad, and **memory-get** to read the best
27523
+ candidates in full. Add only links supported by the notes: same-
27524
+ vault targets go in props.related and natural [[path]] body links; cross-vault targets go in
27525
+ props.related_vault_notes as Vault Name::relative/path.md. Prefer existing People/Projects/system
27526
+ hubs and fewer strong links. Do not trust a background linker to repair a weak write.
27527
+
27528
+ Use the target vault's template, not generic prose. Always set summary, specific reusable tags,
27529
+ source_type, source_ref, related, related_vault_notes, embed, embed_priority, and a valid
27530
+ status; include vault-specific props when relevant. Body patterns: Knowledge = Thesis / Explanation /
27531
+ Examples / Implications; Library = Source Identity / Preserved Material / Extraction Notes; People =
27532
+ Identity / Relationship / Known Context / Interaction History / Open Loops; Projects = Purpose / Current
27533
+ State / Architecture / Open Work / Decisions; Issues = Observation / Impact / Confirmation / Evidence /
27534
+ Resolution; Improvement Log = Improvement / Issue Link / Change Made / Verification / Result. Keep Tasks,
27535
+ Calendar, Communications, Ideas, Inspiration, Experiments, and Sprint aligned to their named purpose.
27536
+ Tags are live vocabulary, not improvised labels: **list-memory-tags** shows what exists and
27537
+ **resolve-memory-tags** returns reuse/create/omit. A new tag is justified only when the concept is central,
27538
+ reusable, and has no exact, alias, or near-equivalent. Use **memory-backlinks**,
27539
+ **memory-graph-universe**, and **memory-graph-path** to trace the linked universe across vaults.
27540
+
27541
+ Scrape deposits are raw evidence and therefore go to **Library** through **library-ingest**, with the full
27542
+ Library template and source metadata. If the source contains durable applicable guidance, create a separate
27543
+ Knowledge companion through prepare-memory-write + memory-capture and link it to the Library source with
27544
+ derived_from; do not replace the raw source with the guide.
27545
+
27546
+ For an update, call **memory-get** first and pass its revision as baseRevision. Preserve the existing
27547
+ title, source, capture time, content context, and props unless the request deliberately changes them.
27548
+ If a legacy backend does not return props, do not overwrite an existing linked note through that surface;
27549
+ report that link preservation cannot be proved. After the write, read it back and search a distinctive
27550
+ phrase to verify persistence and indexing.
27422
27551
  `.trim();
27423
27552
  }
27424
27553
  var SERVER_INSTRUCTIONS;
@@ -28841,7 +28970,7 @@ function localPlanningToolAnnotations(title) {
28841
28970
  }
28842
28971
  function listSavedReports() {
28843
28972
  try {
28844
- const dir = outputBaseDir2();
28973
+ const dir = outputBaseDir();
28845
28974
  return (0, import_node_fs9.readdirSync)(dir).filter((f) => f.endsWith(".md")).map((f) => ({ filename: f, mtimeMs: (0, import_node_fs9.statSync)((0, import_node_path12.join)(dir, f)).mtimeMs })).sort((a, b) => b.mtimeMs - a.mtimeMs).slice(0, 100);
28846
28975
  } catch {
28847
28976
  return [];
@@ -28868,7 +28997,7 @@ function registerSavedReportResources(server) {
28868
28997
  const requested = Array.isArray(variables.filename) ? variables.filename[0] : variables.filename;
28869
28998
  const filename = (0, import_node_path12.basename)(decodeURIComponent(String(requested ?? "")));
28870
28999
  if (!filename.endsWith(".md")) throw new Error("Only saved .md reports can be read");
28871
- const text = (0, import_node_fs9.readFileSync)((0, import_node_path12.join)(outputBaseDir2(), filename), "utf8");
29000
+ const text = (0, import_node_fs9.readFileSync)((0, import_node_path12.join)(outputBaseDir(), filename), "utf8");
28872
29001
  return { contents: [{ uri: uri.href, mimeType: "text/markdown", text }] };
28873
29002
  }
28874
29003
  );
@@ -29467,7 +29596,7 @@ var init_http_mcp_tool_executor = __esm({
29467
29596
  });
29468
29597
 
29469
29598
  // src/services/fanout/export.ts
29470
- function outputBaseDir3() {
29599
+ function outputBaseDir2() {
29471
29600
  return process.env.MCP_SCRAPER_OUTPUT_DIR?.trim() || (0, import_node_path13.join)((0, import_node_os8.homedir)(), "Downloads", "mcp-scraper");
29472
29601
  }
29473
29602
  function safe(value) {
@@ -29478,7 +29607,7 @@ function writeTable(path6, rows, delimiter) {
29478
29607
  }
29479
29608
  function exportFanout(enriched) {
29480
29609
  const stamp = safe(enriched.capturedAt.replace(/[:.]/g, "-"));
29481
- const outputDir = outputBaseDir3();
29610
+ const outputDir = outputBaseDir2();
29482
29611
  const relativeDir = (0, import_node_path13.join)("fanout", `${stamp}-${safe(enriched.platform)}`);
29483
29612
  const dir = (0, import_node_path13.join)(outputDir, relativeDir);
29484
29613
  (0, import_node_fs10.mkdirSync)(dir, { recursive: true });
@@ -30212,10 +30341,10 @@ function ffmpegFilterPath(path6) {
30212
30341
  async function annotateReplayVideo(inputFilePath, outputFilePath, options) {
30213
30342
  if (!options.annotations.length) throw new Error("annotations must include at least one item");
30214
30343
  const size = await videoSize(inputFilePath);
30215
- const tmp = await (0, import_promises10.mkdtemp)((0, import_node_path14.join)((0, import_node_os9.tmpdir)(), "mcp-scraper-ass-"));
30344
+ const tmp = await (0, import_promises9.mkdtemp)((0, import_node_path14.join)((0, import_node_os9.tmpdir)(), "mcp-scraper-ass-"));
30216
30345
  const assPath = (0, import_node_path14.join)(tmp, "annotations.ass");
30217
30346
  try {
30218
- await (0, import_promises10.writeFile)(assPath, buildAssSubtitle(options, size), "utf8");
30347
+ await (0, import_promises9.writeFile)(assPath, buildAssSubtitle(options, size), "utf8");
30219
30348
  await execFileAsync2("ffmpeg", [
30220
30349
  "-y",
30221
30350
  "-i",
@@ -30232,7 +30361,7 @@ async function annotateReplayVideo(inputFilePath, outputFilePath, options) {
30232
30361
  "copy",
30233
30362
  outputFilePath
30234
30363
  ], { maxBuffer: 1024 * 1024 * 20 });
30235
- const out = await (0, import_promises10.stat)(outputFilePath);
30364
+ const out = await (0, import_promises9.stat)(outputFilePath);
30236
30365
  return {
30237
30366
  filePath: outputFilePath,
30238
30367
  bytes: out.size,
@@ -30241,15 +30370,15 @@ async function annotateReplayVideo(inputFilePath, outputFilePath, options) {
30241
30370
  annotationCount: options.annotations.length
30242
30371
  };
30243
30372
  } finally {
30244
- await (0, import_promises10.rm)(tmp, { recursive: true, force: true });
30373
+ await (0, import_promises9.rm)(tmp, { recursive: true, force: true });
30245
30374
  }
30246
30375
  }
30247
- var import_node_child_process4, import_promises10, import_node_os9, import_node_path14, import_node_util2, execFileAsync2;
30376
+ var import_node_child_process4, import_promises9, import_node_os9, import_node_path14, import_node_util2, execFileAsync2;
30248
30377
  var init_replay_annotator = __esm({
30249
30378
  "src/mcp/replay-annotator.ts"() {
30250
30379
  "use strict";
30251
30380
  import_node_child_process4 = require("child_process");
30252
- import_promises10 = require("fs/promises");
30381
+ import_promises9 = require("fs/promises");
30253
30382
  import_node_os9 = require("os");
30254
30383
  import_node_path14 = require("path");
30255
30384
  import_node_util2 = require("util");
@@ -30294,7 +30423,7 @@ function actionResult(tool, sessionId, ok, data, nextRecommendedTool = "browser_
30294
30423
  nextRecommendedTool
30295
30424
  });
30296
30425
  }
30297
- function outputBaseDir4() {
30426
+ function outputBaseDir3() {
30298
30427
  return process.env.MCP_SCRAPER_OUTPUT_DIR?.trim() || (0, import_node_path15.join)((0, import_node_os10.homedir)(), "Downloads", "mcp-scraper");
30299
30428
  }
30300
30429
  function safeFilePart2(value) {
@@ -30304,7 +30433,7 @@ function replayFilePath(sessionId, replayId, filename) {
30304
30433
  const requested = filename?.trim();
30305
30434
  const stamp = (/* @__PURE__ */ new Date()).toISOString().replace(/[:.]/g, "-");
30306
30435
  const name = requested ? safeFilePart2(requested).replace(/\.mp4$/i, "") : `${stamp}-${safeFilePart2(sessionId)}-${safeFilePart2(replayId)}`;
30307
- return (0, import_node_path15.join)(outputBaseDir4(), "browser-replays", `${name}.mp4`);
30436
+ return (0, import_node_path15.join)(outputBaseDir3(), "browser-replays", `${name}.mp4`);
30308
30437
  }
30309
30438
  function slugPart(value) {
30310
30439
  const trimmed = value?.trim().toLowerCase();
@@ -30383,7 +30512,7 @@ function registerBrowserAgentMcpTools(server, opts) {
30383
30512
  }
30384
30513
  const bytes = Buffer.from(await res.arrayBuffer());
30385
30514
  const filePath = replayFilePath(sessionId, replayId, filename);
30386
- (0, import_node_fs11.mkdirSync)((0, import_node_path15.join)(outputBaseDir4(), "browser-replays"), { recursive: true });
30515
+ (0, import_node_fs11.mkdirSync)((0, import_node_path15.join)(outputBaseDir3(), "browser-replays"), { recursive: true });
30387
30516
  (0, import_node_fs11.writeFileSync)(filePath, bytes);
30388
30517
  return {
30389
30518
  ok: true,
@@ -30933,7 +31062,7 @@ function registerBrowserAgentMcpTools(server, opts) {
30933
31062
  try {
30934
31063
  const sourcePath = String(downloaded.data.file_path);
30935
31064
  const outputPath = annotatedReplayFilePath(input.session_id, input.replay_id, input.filename);
30936
- (0, import_node_fs11.mkdirSync)((0, import_node_path15.join)(outputBaseDir4(), "browser-replays"), { recursive: true });
31065
+ (0, import_node_fs11.mkdirSync)((0, import_node_path15.join)(outputBaseDir3(), "browser-replays"), { recursive: true });
30937
31066
  const result = await annotateReplayVideo(sourcePath, outputPath, {
30938
31067
  annotations: input.annotations,
30939
31068
  sourceWidth: input.source_width,
@@ -31078,7 +31207,7 @@ var init_browser_agent_mcp_server = __esm({
31078
31207
  });
31079
31208
 
31080
31209
  // src/mcp/memory-tool-schemas.ts
31081
- var import_zod35, AcceptShareSchema, ApproveSenderSchema, DeclineShareSchema, GetChatLinkSchema, InboxSettingsSchema, InviteAccountSchema, issueKeyTool_scopeShape, IssueKeySchema, ListApprovedSendersSchema, ListKeysSchema, NoteInboxSchema, RemoveApprovedSenderSchema, RevokeChatLinkSchema, RevokeKeySchema, RevokeShareSchema, SetAgentIdentitySchema, SetScopeSchema, ShareNoteSchema, ShareVaultSchema, SwapVaultSchema, SwitchAccountSchema, UnlinkShareSchema, MemoryQuestionsSchema, CreateChannelSchema, GetMessageNoteSchema, ListChannelMembersSchema, ListChannelMessagesSchema, MyMentionsSchema, PollChannelSchema, PostMessageSchema, ReactMessageSchema, RemoveChannelMemberSchema, ReplyMessageSchema, factHistoryTool_entryShape, FactHistorySchema, recordFactTool_factShape, RecordFactSchema, LibraryIngestSchema, DeleteNoteSchema, ExportSchema, GetSchema, ListSchema, notePropsSchema, PutSchema, SearchSchema, SuggestSchema, UploadSchema, TemporalRecallSchema, CreateScheduledActionSchema, DeleteScheduledActionSchema, GetScheduleLinkSchema, GetScheduleStatusSchema, ListScheduledActionsSchema, PauseScheduledActionSchema, ProposeScheduledActionSchema, ResumeScheduledActionSchema, RevokeScheduleLinkSchema, SetScheduleEntitlementSchema, CostUsageSchema, StorageUsageSchema, CreateTableSchema, deleteTableRowsTool_FilterSchema, DeleteTableRowsSchema, DescribeTableSchema, DropTableSchema, InsertTableRowsSchema, ListTablesSchema, queryTableTool_FilterSchema, QueryTableSchema, AddVaultSchema, CreateSecureVaultSchema, DeleteVaultSchema, ListSharedWithMeSchema, ListVaultsSchema, ProvisionDefaultsSchema, VideoAnalyzeStartSchema, VideoAnalyzeStatusSchema, CreateWebhookSchema, ListWebhooksSchema, RevokeWebhookSchema, MEMORY_TOOL_SCHEMAS;
31210
+ var import_zod35, AcceptShareSchema, ApproveSenderSchema, DeclineShareSchema, GetChatLinkSchema, InboxSettingsSchema, InviteAccountSchema, issueKeyTool_scopeShape, IssueKeySchema, ListApprovedSendersSchema, ListKeysSchema, NoteInboxSchema, RemoveApprovedSenderSchema, RevokeChatLinkSchema, RevokeKeySchema, RevokeShareSchema, SetAgentIdentitySchema, SetScopeSchema, ShareNoteSchema, ShareVaultSchema, SwapVaultSchema, SwitchAccountSchema, UnlinkShareSchema, memoryCaptureTool_notePropsSchema, MemoryCaptureSchema, MemoryQuestionsSchema, PrepareMemoryWriteSchema, validateMemoryWriteTool_notePropsSchema, ValidateMemoryWriteSchema, CreateChannelSchema, GetMessageNoteSchema, ListChannelMembersSchema, ListChannelMessagesSchema, MyMentionsSchema, PollChannelSchema, PostMessageSchema, ReactMessageSchema, RemoveChannelMemberSchema, ReplyMessageSchema, factHistoryTool_entryShape, FactHistorySchema, recordFactTool_factShape, RecordFactSchema, NoteBacklinksSchema, GraphPathSchema, GraphUniverseSchema, LibraryIngestSchema, DeleteNoteSchema, ExportSchema, getTool_notePropsSchema, GetSchema, ListSchema, putTool_notePropsSchema, PutSchema, SearchSchema, SuggestSchema, UploadSchema, TemporalRecallSchema, CreateScheduledActionSchema, DeleteScheduledActionSchema, GetScheduleLinkSchema, GetScheduleStatusSchema, ListScheduledActionsSchema, PauseScheduledActionSchema, ProposeScheduledActionSchema, ResumeScheduledActionSchema, RevokeScheduleLinkSchema, SetScheduleEntitlementSchema, CostUsageSchema, StorageUsageSchema, CreateTableSchema, deleteTableRowsTool_FilterSchema, DeleteTableRowsSchema, DescribeTableSchema, DropTableSchema, InsertTableRowsSchema, ListTablesSchema, queryTableTool_FilterSchema, QueryTableSchema, ListTagsSchema, ResolveTagsSchema, UpsertTagSchema, AddVaultSchema, CreateSecureVaultSchema, DeleteVaultSchema, GetVaultContractSchema, ListSharedWithMeSchema, ListVaultsSchema, ProvisionDefaultsSchema, RouteMemorySchema, VideoAnalyzeStartSchema, VideoAnalyzeStatusSchema, CreateWebhookSchema, ListWebhooksSchema, RevokeWebhookSchema, MEMORY_TOOL_SCHEMAS;
31082
31211
  var init_memory_tool_schemas = __esm({
31083
31212
  "src/mcp/memory-tool-schemas.ts"() {
31084
31213
  "use strict";
@@ -31587,6 +31716,52 @@ var init_memory_tool_schemas = __esm({
31587
31716
  openWorldHint: false
31588
31717
  }
31589
31718
  };
31719
+ memoryCaptureTool_notePropsSchema = import_zod35.z.object({
31720
+ status: import_zod35.z.string().describe("Status enum value from the target vault contract."),
31721
+ summary: import_zod35.z.string().describe("Short retrieval-ready description."),
31722
+ tags: import_zod35.z.array(import_zod35.z.string()).describe("AI-generated keyword tags, not vault names."),
31723
+ pinned: import_zod35.z.boolean().describe("Recall boost for important notes."),
31724
+ source_type: import_zod35.z.string().describe("Attribution kind: user, person, url, file, channel, thread, or note."),
31725
+ source_ref: import_zod35.z.string().describe("Attribution reference (URL, path, channel, thread, or source note)."),
31726
+ related: import_zod35.z.array(import_zod35.z.string()).describe("Same-vault links (wiki [[ ]] targets)."),
31727
+ related_vault_notes: import_zod35.z.array(import_zod35.z.string()).describe('Cross-vault references in "Vault Name::relative/path.md" form.'),
31728
+ embed: import_zod35.z.boolean().describe("Whether Smart RAG should index the note."),
31729
+ embed_priority: import_zod35.z.enum(["low", "normal", "high"]).describe("Embedding priority."),
31730
+ embedding_summary: import_zod35.z.string().describe("Optional retrieval-specific summary."),
31731
+ type: import_zod35.z.string().describe("Note type from the target vault contract (also used to route the note)."),
31732
+ domain: import_zod35.z.string().describe("Domain folder for Library/Knowledge (AI, SEO, Copywriting & Ads, Business, Spirituality)."),
31733
+ folder: import_zod35.z.string().describe("Explicit sub-folder within the vault; overrides routing-derived folder."),
31734
+ parentMessageId: import_zod35.z.string().describe("Channel messages only: the path of the top-level message this is a reply to. Absent on top-level messages.")
31735
+ }).partial().passthrough();
31736
+ MemoryCaptureSchema = {
31737
+ id: "memory-capture",
31738
+ upstreamName: "memoryCaptureTool",
31739
+ description: "Strict normal-create path for durable memory. Refuses incomplete notes, writes through memory-put, registers canonical tags, and reads the note back to verify persisted content and props. Call prepare-memory-write first. Reserve memory-put for low-level migrations or deliberate edits.",
31740
+ input: {
31741
+ vault: import_zod35.z.string(),
31742
+ folder: import_zod35.z.string().optional(),
31743
+ path: import_zod35.z.string().min(1),
31744
+ title: import_zod35.z.string().min(1),
31745
+ content: import_zod35.z.string().min(1),
31746
+ props: memoryCaptureTool_notePropsSchema,
31747
+ baseRevision: import_zod35.z.number().optional(),
31748
+ tagDecisions: import_zod35.z.array(import_zod35.z.object({ tag: import_zod35.z.string().min(1), central: import_zod35.z.boolean(), reusable: import_zod35.z.boolean(), description: import_zod35.z.string().optional() })).max(8).optional().describe("Required justification for any tag that does not already exist. Existing exact/alias/near tags are canonicalized automatically; a new tag is accepted only when its matching decision has central=true and reusable=true.")
31749
+ },
31750
+ output: {
31751
+ ok: import_zod35.z.boolean(),
31752
+ valid: import_zod35.z.boolean().optional(),
31753
+ errors: import_zod35.z.array(import_zod35.z.string()).optional(),
31754
+ warnings: import_zod35.z.array(import_zod35.z.string()).optional(),
31755
+ tagResolutions: import_zod35.z.array(import_zod35.z.object({ candidate: import_zod35.z.string(), action: import_zod35.z.enum(["reuse", "create", "omit"]), tag: import_zod35.z.string().optional(), reason: import_zod35.z.string() })).optional(),
31756
+ note: import_zod35.z.object({ path: import_zod35.z.string(), title: import_zod35.z.string(), updatedAt: import_zod35.z.string(), revision: import_zod35.z.number() }).optional(),
31757
+ indexed: import_zod35.z.number().optional(),
31758
+ verified: import_zod35.z.object({ contentBytes: import_zod35.z.number(), propsPersisted: import_zod35.z.boolean(), revision: import_zod35.z.number() }).optional(),
31759
+ conflict: import_zod35.z.boolean().optional(),
31760
+ code: import_zod35.z.string().optional(),
31761
+ error: import_zod35.z.string().optional()
31762
+ },
31763
+ annotations: { title: "Capture Governed Memory", readOnlyHint: false, destructiveHint: false, idempotentHint: true, openWorldHint: true }
31764
+ };
31590
31765
  MemoryQuestionsSchema = {
31591
31766
  id: "memory-questions",
31592
31767
  upstreamName: "memoryQuestionsTool",
@@ -31616,6 +31791,67 @@ var init_memory_tool_schemas = __esm({
31616
31791
  openWorldHint: true
31617
31792
  }
31618
31793
  };
31794
+ PrepareMemoryWriteSchema = {
31795
+ id: "prepare-memory-write",
31796
+ upstreamName: "prepareMemoryWriteTool",
31797
+ description: "Mandatory planning pass for a normal new memory. Routes the note, returns the live template and natural vault relationships, resolves proposed tags against the registry, and shortlists interlink opportunities that must be read/reviewed before capture.",
31798
+ input: {
31799
+ title: import_zod35.z.string().min(1),
31800
+ content: import_zod35.z.string().min(1),
31801
+ source: import_zod35.z.string().optional(),
31802
+ type: import_zod35.z.string().optional(),
31803
+ vault: import_zod35.z.string().optional(),
31804
+ tagCandidates: import_zod35.z.array(import_zod35.z.object({ tag: import_zod35.z.string().min(1), central: import_zod35.z.boolean().optional(), reusable: import_zod35.z.boolean().optional(), description: import_zod35.z.string().optional() })).max(20).optional(),
31805
+ maxLinks: import_zod35.z.number().int().min(1).max(20).optional()
31806
+ },
31807
+ output: {
31808
+ ok: import_zod35.z.boolean(),
31809
+ route: import_zod35.z.record(import_zod35.z.unknown()).optional(),
31810
+ contract: import_zod35.z.record(import_zod35.z.unknown()).optional(),
31811
+ tagResolutions: import_zod35.z.array(import_zod35.z.record(import_zod35.z.unknown())).optional(),
31812
+ linkOpportunities: import_zod35.z.array(import_zod35.z.record(import_zod35.z.unknown())).optional(),
31813
+ instructions: import_zod35.z.array(import_zod35.z.string()).optional(),
31814
+ error: import_zod35.z.string().optional()
31815
+ },
31816
+ annotations: { title: "Prepare Memory Write", readOnlyHint: true, destructiveHint: false, idempotentHint: true, openWorldHint: false }
31817
+ };
31818
+ validateMemoryWriteTool_notePropsSchema = import_zod35.z.object({
31819
+ status: import_zod35.z.string().describe("Status enum value from the target vault contract."),
31820
+ summary: import_zod35.z.string().describe("Short retrieval-ready description."),
31821
+ tags: import_zod35.z.array(import_zod35.z.string()).describe("AI-generated keyword tags, not vault names."),
31822
+ pinned: import_zod35.z.boolean().describe("Recall boost for important notes."),
31823
+ source_type: import_zod35.z.string().describe("Attribution kind: user, person, url, file, channel, thread, or note."),
31824
+ source_ref: import_zod35.z.string().describe("Attribution reference (URL, path, channel, thread, or source note)."),
31825
+ related: import_zod35.z.array(import_zod35.z.string()).describe("Same-vault links (wiki [[ ]] targets)."),
31826
+ related_vault_notes: import_zod35.z.array(import_zod35.z.string()).describe('Cross-vault references in "Vault Name::relative/path.md" form.'),
31827
+ embed: import_zod35.z.boolean().describe("Whether Smart RAG should index the note."),
31828
+ embed_priority: import_zod35.z.enum(["low", "normal", "high"]).describe("Embedding priority."),
31829
+ embedding_summary: import_zod35.z.string().describe("Optional retrieval-specific summary."),
31830
+ type: import_zod35.z.string().describe("Note type from the target vault contract (also used to route the note)."),
31831
+ domain: import_zod35.z.string().describe("Domain folder for Library/Knowledge (AI, SEO, Copywriting & Ads, Business, Spirituality)."),
31832
+ folder: import_zod35.z.string().describe("Explicit sub-folder within the vault; overrides routing-derived folder."),
31833
+ parentMessageId: import_zod35.z.string().describe("Channel messages only: the path of the top-level message this is a reply to. Absent on top-level messages.")
31834
+ }).partial().passthrough();
31835
+ ValidateMemoryWriteSchema = {
31836
+ id: "validate-memory-write",
31837
+ upstreamName: "validateMemoryWriteTool",
31838
+ description: "Validate a proposed governed note without writing it. Checks template completeness, vault status/type, canonical tag count, attribution, Obsidian link primitives, and retrieval metadata.",
31839
+ input: {
31840
+ vault: import_zod35.z.string(),
31841
+ title: import_zod35.z.string(),
31842
+ content: import_zod35.z.string(),
31843
+ props: validateMemoryWriteTool_notePropsSchema
31844
+ },
31845
+ output: {
31846
+ ok: import_zod35.z.boolean(),
31847
+ valid: import_zod35.z.boolean().optional(),
31848
+ errors: import_zod35.z.array(import_zod35.z.string()).optional(),
31849
+ warnings: import_zod35.z.array(import_zod35.z.string()).optional(),
31850
+ normalizedTags: import_zod35.z.array(import_zod35.z.string()).optional(),
31851
+ error: import_zod35.z.string().optional()
31852
+ },
31853
+ annotations: { title: "Validate Memory Write", readOnlyHint: true, destructiveHint: false, idempotentHint: true, openWorldHint: false }
31854
+ };
31619
31855
  CreateChannelSchema = {
31620
31856
  id: "create-channel",
31621
31857
  upstreamName: "createChannelTool",
@@ -31956,18 +32192,78 @@ var init_memory_tool_schemas = __esm({
31956
32192
  openWorldHint: false
31957
32193
  }
31958
32194
  };
32195
+ NoteBacklinksSchema = {
32196
+ id: "memory-backlinks",
32197
+ upstreamName: "noteBacklinksTool",
32198
+ description: "Return every resolved incoming link to a note across all accessible vaults, including typed metadata links and Obsidian wiki links.",
32199
+ input: {
32200
+ note: import_zod35.z.string().min(1),
32201
+ vault: import_zod35.z.string().optional()
32202
+ },
32203
+ output: {
32204
+ ok: import_zod35.z.boolean(),
32205
+ nodes: import_zod35.z.array(import_zod35.z.object({ id: import_zod35.z.string(), vault: import_zod35.z.string(), path: import_zod35.z.string(), title: import_zod35.z.string(), resolved: import_zod35.z.boolean() })).optional(),
32206
+ edges: import_zod35.z.array(import_zod35.z.object({ source: import_zod35.z.string(), target: import_zod35.z.string(), type: import_zod35.z.string(), evidence: import_zod35.z.string(), sourceField: import_zod35.z.string(), resolved: import_zod35.z.boolean() })).optional(),
32207
+ error: import_zod35.z.string().optional()
32208
+ },
32209
+ annotations: { title: "Memory Backlinks", readOnlyHint: true, destructiveHint: false, idempotentHint: true, openWorldHint: false }
32210
+ };
32211
+ GraphPathSchema = {
32212
+ id: "memory-graph-path",
32213
+ upstreamName: "graphPathTool",
32214
+ description: "Find the shortest navigable connection between two memory notes across vault boundaries. Returns an empty graph when no path exists within maxDepth.",
32215
+ input: {
32216
+ from: import_zod35.z.string().min(1),
32217
+ to: import_zod35.z.string().min(1),
32218
+ fromVault: import_zod35.z.string().optional(),
32219
+ toVault: import_zod35.z.string().optional(),
32220
+ maxDepth: import_zod35.z.number().int().min(1).max(12).optional()
32221
+ },
32222
+ output: {
32223
+ ok: import_zod35.z.boolean(),
32224
+ nodes: import_zod35.z.array(import_zod35.z.object({ id: import_zod35.z.string(), vault: import_zod35.z.string(), path: import_zod35.z.string(), title: import_zod35.z.string(), resolved: import_zod35.z.boolean() })).optional(),
32225
+ edges: import_zod35.z.array(import_zod35.z.object({ source: import_zod35.z.string(), target: import_zod35.z.string(), type: import_zod35.z.string(), evidence: import_zod35.z.string(), sourceField: import_zod35.z.string(), resolved: import_zod35.z.boolean() })).optional(),
32226
+ error: import_zod35.z.string().optional()
32227
+ },
32228
+ annotations: { title: "Memory Graph Path", readOnlyHint: true, destructiveHint: false, idempotentHint: true, openWorldHint: false }
32229
+ };
32230
+ GraphUniverseSchema = {
32231
+ id: "memory-graph-universe",
32232
+ upstreamName: "graphUniverseTool",
32233
+ description: "Traverse the interlinked memory universe around one note in every direction, like an Obsidian local graph. Returns nodes and typed edges to a bounded depth.",
32234
+ input: {
32235
+ note: import_zod35.z.string().min(1),
32236
+ vault: import_zod35.z.string().optional(),
32237
+ depth: import_zod35.z.number().int().min(0).max(6).optional(),
32238
+ maxNodes: import_zod35.z.number().int().min(1).max(500).optional()
32239
+ },
32240
+ output: {
32241
+ ok: import_zod35.z.boolean(),
32242
+ root: import_zod35.z.string().optional(),
32243
+ depth: import_zod35.z.number().optional(),
32244
+ truncated: import_zod35.z.boolean().optional(),
32245
+ nodes: import_zod35.z.array(import_zod35.z.object({ id: import_zod35.z.string(), vault: import_zod35.z.string(), path: import_zod35.z.string(), title: import_zod35.z.string(), resolved: import_zod35.z.boolean() })).optional(),
32246
+ edges: import_zod35.z.array(import_zod35.z.object({ source: import_zod35.z.string(), target: import_zod35.z.string(), type: import_zod35.z.string(), evidence: import_zod35.z.string(), sourceField: import_zod35.z.string(), resolved: import_zod35.z.boolean() })).optional(),
32247
+ error: import_zod35.z.string().optional()
32248
+ },
32249
+ annotations: { title: "Memory Graph Universe", readOnlyHint: true, destructiveHint: false, idempotentHint: true, openWorldHint: false }
32250
+ };
31959
32251
  LibraryIngestSchema = {
31960
32252
  id: "library-ingest",
31961
32253
  upstreamName: "libraryIngestTool",
31962
32254
  description: "Deposit a scrape, transcript, or generated output into the tenant Library vault for later semantic recall. Content is embedded for per-tenant search and best-effort mirrored to a local vault when configured. Requires write scope on the target vault.",
31963
32255
  input: {
31964
32256
  vault: import_zod35.z.string().optional().describe(
31965
- 'Vault to deposit into. Optional; defaults to the session active vault, then the first entitled vault, then "Library".'
32257
+ "Vault to deposit into. Optional and normally omitted: raw scrapes always default to Library. Override only for a deliberate nonstandard migration."
31966
32258
  ),
31967
32259
  title: import_zod35.z.string().min(1).describe("Short human-readable title for the item; used to build the stored path. Must be non-empty."),
31968
32260
  content: import_zod35.z.string().min(1).describe("The full captured text (scrape/transcript/output) to store and index. Must be non-empty."),
31969
32261
  source: import_zod35.z.string().min(1).describe("Provenance of the content, e.g. a URL or tool name. Must be non-empty."),
31970
32262
  capturedAt: import_zod35.z.string().optional().describe("ISO-8601 capture timestamp. Optional; defaults to now. Also seeds the deterministic storage path."),
32263
+ summary: import_zod35.z.string().optional().describe("Retrieval-ready source summary. Optional; a provenance summary is generated when omitted."),
32264
+ tags: import_zod35.z.array(import_zod35.z.string()).max(8).optional().describe("Reviewed canonical tags. Existing tags should be resolved first; when omitted, deterministic source/topic tags are generated."),
32265
+ related: import_zod35.z.array(import_zod35.z.string()).optional().describe("Reviewed same-vault Library note paths."),
32266
+ relatedVaultNotes: import_zod35.z.array(import_zod35.z.string()).optional().describe("Reviewed cross-vault references in Vault::path.md form."),
31971
32267
  localVaultPath: import_zod35.z.string().optional().describe("Filesystem root to also mirror the item to. Optional; falls back to MEMORY_LOCAL_VAULT_ROOT env when set.")
31972
32268
  },
31973
32269
  output: {
@@ -31977,6 +32273,7 @@ var init_memory_tool_schemas = __esm({
31977
32273
  path: import_zod35.z.string().optional().describe("Vault-relative path the item was stored at (under library/...)."),
31978
32274
  indexed: import_zod35.z.number().optional().describe("Number of search chunks indexed (0 if embedding failed but the note still saved)."),
31979
32275
  dualWritten: import_zod35.z.boolean().optional().describe("True if the item was also mirrored to the local filesystem vault."),
32276
+ nextStep: import_zod35.z.string().optional().describe("Recommended extraction action after the raw Library source is safe."),
31980
32277
  code: import_zod35.z.string().optional().describe("Machine-readable denial code when ok is false: quota_exceeded or free_cost_cap."),
31981
32278
  error: import_zod35.z.string().optional().describe("Human-readable failure reason when ok is false.")
31982
32279
  },
@@ -32054,10 +32351,27 @@ var init_memory_tool_schemas = __esm({
32054
32351
  openWorldHint: false
32055
32352
  }
32056
32353
  };
32354
+ getTool_notePropsSchema = import_zod35.z.object({
32355
+ status: import_zod35.z.string().describe("Status enum value from the target vault contract."),
32356
+ summary: import_zod35.z.string().describe("Short retrieval-ready description."),
32357
+ tags: import_zod35.z.array(import_zod35.z.string()).describe("AI-generated keyword tags, not vault names."),
32358
+ pinned: import_zod35.z.boolean().describe("Recall boost for important notes."),
32359
+ source_type: import_zod35.z.string().describe("Attribution kind: user, person, url, file, channel, thread, or note."),
32360
+ source_ref: import_zod35.z.string().describe("Attribution reference (URL, path, channel, thread, or source note)."),
32361
+ related: import_zod35.z.array(import_zod35.z.string()).describe("Same-vault links (wiki [[ ]] targets)."),
32362
+ related_vault_notes: import_zod35.z.array(import_zod35.z.string()).describe('Cross-vault references in "Vault Name::relative/path.md" form.'),
32363
+ embed: import_zod35.z.boolean().describe("Whether Smart RAG should index the note."),
32364
+ embed_priority: import_zod35.z.enum(["low", "normal", "high"]).describe("Embedding priority."),
32365
+ embedding_summary: import_zod35.z.string().describe("Optional retrieval-specific summary."),
32366
+ type: import_zod35.z.string().describe("Note type from the target vault contract (also used to route the note)."),
32367
+ domain: import_zod35.z.string().describe("Domain folder for Library/Knowledge (AI, SEO, Copywriting & Ads, Business, Spirituality)."),
32368
+ folder: import_zod35.z.string().describe("Explicit sub-folder within the vault; overrides routing-derived folder."),
32369
+ parentMessageId: import_zod35.z.string().describe("Channel messages only: the path of the top-level message this is a reply to. Absent on top-level messages.")
32370
+ }).partial().passthrough();
32057
32371
  GetSchema = {
32058
32372
  id: "memory-get",
32059
32373
  upstreamName: "getTool",
32060
- description: "Read a single note from a vault by its exact path, or by shareId for a note shared with you and accepted. Returns a revision number \u2014 pass it as baseRevision on a later memory-put/delete-note to detect a concurrent edit instead of silently overwriting it. Requires read scope.",
32374
+ description: "Read a single note from a vault by its exact path, or by shareId for a note shared with you and accepted. Owned notes include their stored Obsidian props so edits can preserve links and template metadata. Returns a revision number \u2014 pass it as baseRevision on a later memory-put/delete-note to detect a concurrent edit instead of silently overwriting it. Requires read scope.",
32061
32375
  input: {
32062
32376
  vault: import_zod35.z.string().optional().describe(
32063
32377
  "Vault to read from. Optional; defaults to the session active vault, then the first vault the caller is entitled to. Ignored when shareId is given."
@@ -32077,6 +32391,7 @@ var init_memory_tool_schemas = __esm({
32077
32391
  capturedAt: import_zod35.z.string().describe("ISO-8601 timestamp of when the note was first captured."),
32078
32392
  revision: import_zod35.z.number().describe("Current revision number. Pass this as baseRevision when editing, so a concurrent edit by someone else is detected instead of silently overwritten."),
32079
32393
  updatedBy: import_zod35.z.string().nullable().describe("Identity that made the last write, if known."),
32394
+ props: getTool_notePropsSchema.optional().describe("Stored Obsidian metadata for an owned note, including links and vault-specific template fields. Omitted for secure or individually shared notes."),
32080
32395
  sharedBy: import_zod35.z.string().optional().describe("Present only when read via shareId: the identity who owns and shared this note."),
32081
32396
  othersEditing: import_zod35.z.array(import_zod35.z.string()).optional().describe("Advisory only, not a lock: identities who read this same note within the last ~2 minutes and may be about to edit it. Worth a heads-up to the human before you write; the real safety net is still baseRevision on the write itself.")
32082
32397
  }).optional().describe("The full note. Present when ok is true."),
@@ -32124,7 +32439,7 @@ var init_memory_tool_schemas = __esm({
32124
32439
  openWorldHint: false
32125
32440
  }
32126
32441
  };
32127
- notePropsSchema = import_zod35.z.object({
32442
+ putTool_notePropsSchema = import_zod35.z.object({
32128
32443
  status: import_zod35.z.string().describe("Status enum value from the target vault contract."),
32129
32444
  summary: import_zod35.z.string().describe("Short retrieval-ready description."),
32130
32445
  tags: import_zod35.z.array(import_zod35.z.string()).describe("AI-generated keyword tags, not vault names."),
@@ -32140,7 +32455,7 @@ var init_memory_tool_schemas = __esm({
32140
32455
  domain: import_zod35.z.string().describe("Domain folder for Library/Knowledge (AI, SEO, Copywriting & Ads, Business, Spirituality)."),
32141
32456
  folder: import_zod35.z.string().describe("Explicit sub-folder within the vault; overrides routing-derived folder."),
32142
32457
  parentMessageId: import_zod35.z.string().describe("Channel messages only: the path of the top-level message this is a reply to. Absent on top-level messages.")
32143
- }).partial();
32458
+ }).partial().passthrough();
32144
32459
  PutSchema = {
32145
32460
  id: "memory-put",
32146
32461
  upstreamName: "putTool",
@@ -32153,7 +32468,7 @@ var init_memory_tool_schemas = __esm({
32153
32468
  shareId: import_zod35.z.string().optional().describe("Edit a note someone individually shared with you and you accepted (accept-share), by its shareId, instead of vault+path. Requires the share to grant edit permission, and baseRevision is mandatory (get the current revision first) since you are editing alongside the owner and possibly others."),
32154
32469
  title: import_zod35.z.string().optional().describe("Optional human-readable title; defaults are derived from the path when omitted."),
32155
32470
  content: import_zod35.z.string().min(1).describe("The full note body to store and index for semantic search. Must be non-empty."),
32156
- props: notePropsSchema.optional().describe("Obsidian note primitives (status, summary, tags, pinned, source_type, source_ref, related, related_vault_notes, embed, embed_priority, type, domain, folder). Stored on the note so it round-trips the vault shape; type/domain/folder also steer routing when no vault is given."),
32471
+ props: putTool_notePropsSchema.optional().describe("Obsidian note primitives plus vault-specific template fields. On edits, supplied fields patch the stored props instead of replacing the whole object; pass an empty array to deliberately clear a link list. Type/domain/folder also steer routing when no vault is given."),
32157
32472
  baseRevision: import_zod35.z.number().optional().describe("Revision the edit is based on (from a prior get/put). When provided, the write only applies if the note is still at this revision; otherwise it is rejected as a conflict instead of silently overwriting a concurrent edit. Omit for last-write-wins (fine for solo notes).")
32158
32473
  },
32159
32474
  output: {
@@ -32767,6 +33082,51 @@ var init_memory_tool_schemas = __esm({
32767
33082
  openWorldHint: false
32768
33083
  }
32769
33084
  };
33085
+ ListTagsSchema = {
33086
+ id: "list-memory-tags",
33087
+ upstreamName: "listTagsTool",
33088
+ description: "List the live canonical tag vocabulary, aliases, usage counts, and per-vault distribution. Use this before choosing tags so existing concepts are reused instead of fragmented.",
33089
+ input: {
33090
+ includeDeprecated: import_zod35.z.boolean().optional()
33091
+ },
33092
+ output: {
33093
+ ok: import_zod35.z.boolean(),
33094
+ tags: import_zod35.z.array(import_zod35.z.object({ tag: import_zod35.z.string(), description: import_zod35.z.string().nullable(), aliases: import_zod35.z.array(import_zod35.z.string()), status: import_zod35.z.enum(["active", "deprecated"]), usageCount: import_zod35.z.number(), vaultUsage: import_zod35.z.record(import_zod35.z.number()) })).optional(),
33095
+ error: import_zod35.z.string().optional()
33096
+ },
33097
+ annotations: { title: "List Memory Tags", readOnlyHint: true, destructiveHint: false, idempotentHint: true, openWorldHint: false }
33098
+ };
33099
+ ResolveTagsSchema = {
33100
+ id: "resolve-memory-tags",
33101
+ upstreamName: "resolveTagsTool",
33102
+ description: "Resolve proposed concepts against the live tag vocabulary. Returns reuse, create, or omit. A new tag is allowed only when no equivalent exists and the caller marks the concept central and reusable.",
33103
+ input: {
33104
+ candidates: import_zod35.z.array(import_zod35.z.object({ tag: import_zod35.z.string().min(1), central: import_zod35.z.boolean().optional(), reusable: import_zod35.z.boolean().optional(), description: import_zod35.z.string().optional() })).min(1).max(20)
33105
+ },
33106
+ output: {
33107
+ ok: import_zod35.z.boolean(),
33108
+ resolutions: import_zod35.z.array(import_zod35.z.object({ candidate: import_zod35.z.string(), normalized: import_zod35.z.string(), action: import_zod35.z.enum(["reuse", "create", "omit"]), tag: import_zod35.z.string().optional(), matchedBy: import_zod35.z.enum(["exact", "alias", "near"]).optional(), score: import_zod35.z.number().optional(), reason: import_zod35.z.string() })).optional(),
33109
+ error: import_zod35.z.string().optional()
33110
+ },
33111
+ annotations: { title: "Resolve Memory Tags", readOnlyHint: true, destructiveHint: false, idempotentHint: true, openWorldHint: false }
33112
+ };
33113
+ UpsertTagSchema = {
33114
+ id: "upsert-memory-tag",
33115
+ upstreamName: "upsertTagTool",
33116
+ description: "Define or curate one canonical tag, its meaning, aliases, and lifecycle. Use only after resolve-memory-tags returns create, or to merge/deprecate vocabulary intentionally. Requires write scope.",
33117
+ input: {
33118
+ tag: import_zod35.z.string().min(1),
33119
+ description: import_zod35.z.string().optional(),
33120
+ aliases: import_zod35.z.array(import_zod35.z.string()).optional(),
33121
+ status: import_zod35.z.enum(["active", "deprecated"]).optional()
33122
+ },
33123
+ output: {
33124
+ ok: import_zod35.z.boolean(),
33125
+ tag: import_zod35.z.object({ tag: import_zod35.z.string(), description: import_zod35.z.string().nullable(), aliases: import_zod35.z.array(import_zod35.z.string()), status: import_zod35.z.enum(["active", "deprecated"]) }).optional(),
33126
+ error: import_zod35.z.string().optional()
33127
+ },
33128
+ annotations: { title: "Upsert Memory Tag", readOnlyHint: false, destructiveHint: false, idempotentHint: true, openWorldHint: false }
33129
+ };
32770
33130
  AddVaultSchema = {
32771
33131
  id: "add-vault",
32772
33132
  upstreamName: "addVaultTool",
@@ -32834,6 +33194,21 @@ var init_memory_tool_schemas = __esm({
32834
33194
  openWorldHint: false
32835
33195
  }
32836
33196
  };
33197
+ GetVaultContractSchema = {
33198
+ id: "get-vault-contract",
33199
+ upstreamName: "getVaultContractTool",
33200
+ description: "Read the machine-enforced purpose, template, statuses, types, natural neighbor vaults, and typed relationship guidance for one of the 13 governed Obsidian-style vaults. Call before composing a note when the correct shape is uncertain.",
33201
+ input: {
33202
+ vault: import_zod35.z.string().describe("One governed vault: Ideas, Inspiration, Knowledge, Library, People, Communications, Calendar, Tasks, Projects, Issues, Improvement Log, Experiments, or Sprint.")
33203
+ },
33204
+ output: {
33205
+ ok: import_zod35.z.boolean(),
33206
+ vault: import_zod35.z.string().optional(),
33207
+ contract: import_zod35.z.record(import_zod35.z.unknown()).optional(),
33208
+ error: import_zod35.z.string().optional()
33209
+ },
33210
+ annotations: { title: "Get Vault Contract", readOnlyHint: true, destructiveHint: false, idempotentHint: true, openWorldHint: false }
33211
+ };
32837
33212
  ListSharedWithMeSchema = {
32838
33213
  id: "list-shared-with-me",
32839
33214
  upstreamName: "listSharedWithMeTool",
@@ -32918,6 +33293,26 @@ var init_memory_tool_schemas = __esm({
32918
33293
  openWorldHint: false
32919
33294
  }
32920
33295
  };
33296
+ RouteMemorySchema = {
33297
+ id: "route-memory",
33298
+ upstreamName: "routeMemoryTool",
33299
+ description: "Choose the correct governed vault and folder from content intent. Raw scraped/source material routes to Library; distilled applicable guidance routes to Knowledge. Returns the live vault contract so the result does not depend on prompt memory.",
33300
+ input: {
33301
+ title: import_zod35.z.string().min(1),
33302
+ content: import_zod35.z.string().min(1),
33303
+ type: import_zod35.z.string().optional(),
33304
+ source: import_zod35.z.string().optional()
33305
+ },
33306
+ output: {
33307
+ ok: import_zod35.z.boolean(),
33308
+ vault: import_zod35.z.string().optional(),
33309
+ folder: import_zod35.z.string().optional(),
33310
+ reason: import_zod35.z.string().optional(),
33311
+ contract: import_zod35.z.record(import_zod35.z.unknown()).optional(),
33312
+ error: import_zod35.z.string().optional()
33313
+ },
33314
+ annotations: { title: "Route Memory", readOnlyHint: true, destructiveHint: false, idempotentHint: true, openWorldHint: false }
33315
+ };
32921
33316
  VideoAnalyzeStartSchema = {
32922
33317
  id: "video-analyze-start",
32923
33318
  upstreamName: "videoAnalyzeStartTool",
@@ -33040,7 +33435,10 @@ var init_memory_tool_schemas = __esm({
33040
33435
  SwapVaultSchema,
33041
33436
  SwitchAccountSchema,
33042
33437
  UnlinkShareSchema,
33438
+ MemoryCaptureSchema,
33043
33439
  MemoryQuestionsSchema,
33440
+ PrepareMemoryWriteSchema,
33441
+ ValidateMemoryWriteSchema,
33044
33442
  CreateChannelSchema,
33045
33443
  GetMessageNoteSchema,
33046
33444
  ListChannelMembersSchema,
@@ -33053,6 +33451,9 @@ var init_memory_tool_schemas = __esm({
33053
33451
  ReplyMessageSchema,
33054
33452
  FactHistorySchema,
33055
33453
  RecordFactSchema,
33454
+ NoteBacklinksSchema,
33455
+ GraphPathSchema,
33456
+ GraphUniverseSchema,
33056
33457
  LibraryIngestSchema,
33057
33458
  DeleteNoteSchema,
33058
33459
  ExportSchema,
@@ -33082,12 +33483,17 @@ var init_memory_tool_schemas = __esm({
33082
33483
  InsertTableRowsSchema,
33083
33484
  ListTablesSchema,
33084
33485
  QueryTableSchema,
33486
+ ListTagsSchema,
33487
+ ResolveTagsSchema,
33488
+ UpsertTagSchema,
33085
33489
  AddVaultSchema,
33086
33490
  CreateSecureVaultSchema,
33087
33491
  DeleteVaultSchema,
33492
+ GetVaultContractSchema,
33088
33493
  ListSharedWithMeSchema,
33089
33494
  ListVaultsSchema,
33090
33495
  ProvisionDefaultsSchema,
33496
+ RouteMemorySchema,
33091
33497
  VideoAnalyzeStartSchema,
33092
33498
  VideoAnalyzeStatusSchema,
33093
33499
  CreateWebhookSchema,
@@ -33295,8 +33701,44 @@ function registerReadCutovers() {
33295
33701
  const result = await import_list_webhooks.listWebhooksTool.execute({ ...input, apiKey }, {});
33296
33702
  return { content: [{ type: "text", text: JSON.stringify(result ?? {}) }], structuredContent: result ?? {}, isError: false };
33297
33703
  });
33704
+ registerCutover("getVaultContractTool", async (input, apiKey) => {
33705
+ const result = await import_get_vault_contract.getVaultContractTool.execute({ ...input, apiKey }, {});
33706
+ return { content: [{ type: "text", text: JSON.stringify(result ?? {}) }], structuredContent: result ?? {}, isError: false };
33707
+ });
33708
+ registerCutover("routeMemoryTool", async (input, apiKey) => {
33709
+ const result = await import_route_memory.routeMemoryTool.execute({ ...input, apiKey }, {});
33710
+ return { content: [{ type: "text", text: JSON.stringify(result ?? {}) }], structuredContent: result ?? {}, isError: false };
33711
+ });
33712
+ registerCutover("listTagsTool", async (input, apiKey) => {
33713
+ const result = await import_list_tags.listTagsTool.execute({ ...input, apiKey }, {});
33714
+ return { content: [{ type: "text", text: JSON.stringify(result ?? {}) }], structuredContent: result ?? {}, isError: false };
33715
+ });
33716
+ registerCutover("resolveTagsTool", async (input, apiKey) => {
33717
+ const result = await import_resolve_tags.resolveTagsTool.execute({ ...input, apiKey }, {});
33718
+ return { content: [{ type: "text", text: JSON.stringify(result ?? {}) }], structuredContent: result ?? {}, isError: false };
33719
+ });
33720
+ registerCutover("noteBacklinksTool", async (input, apiKey) => {
33721
+ const result = await import_memory_backlinks.noteBacklinksTool.execute({ ...input, apiKey }, {});
33722
+ return { content: [{ type: "text", text: JSON.stringify(result ?? {}) }], structuredContent: result ?? {}, isError: false };
33723
+ });
33724
+ registerCutover("graphUniverseTool", async (input, apiKey) => {
33725
+ const result = await import_memory_graph_universe.graphUniverseTool.execute({ ...input, apiKey }, {});
33726
+ return { content: [{ type: "text", text: JSON.stringify(result ?? {}) }], structuredContent: result ?? {}, isError: false };
33727
+ });
33728
+ registerCutover("graphPathTool", async (input, apiKey) => {
33729
+ const result = await import_memory_graph_path.graphPathTool.execute({ ...input, apiKey }, {});
33730
+ return { content: [{ type: "text", text: JSON.stringify(result ?? {}) }], structuredContent: result ?? {}, isError: false };
33731
+ });
33732
+ registerCutover("prepareMemoryWriteTool", async (input, apiKey) => {
33733
+ const result = await import_prepare_memory_write.prepareMemoryWriteTool.execute({ ...input, apiKey }, {});
33734
+ return { content: [{ type: "text", text: JSON.stringify(result ?? {}) }], structuredContent: result ?? {}, isError: false };
33735
+ });
33736
+ registerCutover("validateMemoryWriteTool", async (input, apiKey) => {
33737
+ const result = await import_validate_memory_write.validateMemoryWriteTool.execute({ ...input, apiKey }, {});
33738
+ return { content: [{ type: "text", text: JSON.stringify(result ?? {}) }], structuredContent: result ?? {}, isError: false };
33739
+ });
33298
33740
  }
33299
- var import_list_approved_senders, import_note_inbox, import_get_message_note, import_list_channel_members, import_my_mentions, import_fact_history, import_export2, import_get, import_list, import_search, import_suggest, import_temporal_recall, import_list_scheduled_actions, import_cost_usage, import_storage_usage, import_table_describe, import_table_list, import_table_query, import_list_shared_with_me, import_list_vaults, import_status, import_list_webhooks, registered;
33741
+ var import_list_approved_senders, import_note_inbox, import_get_message_note, import_list_channel_members, import_my_mentions, import_fact_history, import_export2, import_get, import_list, import_search, import_suggest, import_temporal_recall, import_list_scheduled_actions, import_cost_usage, import_storage_usage, import_table_describe, import_table_list, import_table_query, import_list_shared_with_me, import_list_vaults, import_status, import_list_webhooks, import_get_vault_contract, import_route_memory, import_list_tags, import_resolve_tags, import_memory_backlinks, import_memory_graph_universe, import_memory_graph_path, import_prepare_memory_write, import_validate_memory_write, registered;
33300
33742
  var init_reads = __esm({
33301
33743
  "src/mcp/memory-cutover/reads.ts"() {
33302
33744
  "use strict";
@@ -33323,6 +33765,15 @@ var init_reads = __esm({
33323
33765
  import_list_vaults = require("mcpscraper-memory-tools/tools/vaults/list-vaults");
33324
33766
  import_status = require("mcpscraper-memory-tools/tools/video/status");
33325
33767
  import_list_webhooks = require("mcpscraper-memory-tools/tools/webhooks/list-webhooks");
33768
+ import_get_vault_contract = require("mcpscraper-memory-tools/tools/vaults/get-vault-contract");
33769
+ import_route_memory = require("mcpscraper-memory-tools/tools/vaults/route-memory");
33770
+ import_list_tags = require("mcpscraper-memory-tools/tools/tags/list-tags");
33771
+ import_resolve_tags = require("mcpscraper-memory-tools/tools/tags/resolve-tags");
33772
+ import_memory_backlinks = require("mcpscraper-memory-tools/tools/graph/memory-backlinks");
33773
+ import_memory_graph_universe = require("mcpscraper-memory-tools/tools/graph/memory-graph-universe");
33774
+ import_memory_graph_path = require("mcpscraper-memory-tools/tools/graph/memory-graph-path");
33775
+ import_prepare_memory_write = require("mcpscraper-memory-tools/tools/capture/prepare-memory-write");
33776
+ import_validate_memory_write = require("mcpscraper-memory-tools/tools/capture/validate-memory-write");
33326
33777
  registered = false;
33327
33778
  }
33328
33779
  });
@@ -33467,8 +33918,16 @@ function registerWriteCutovers() {
33467
33918
  const result = await import_delete_vault.deleteVaultTool.execute({ ...input, apiKey }, {});
33468
33919
  return { content: [{ type: "text", text: JSON.stringify(result ?? {}) }], structuredContent: result ?? {}, isError: false };
33469
33920
  });
33921
+ registerCutover("memoryCaptureTool", async (input, apiKey) => {
33922
+ const result = await import_memory_capture.memoryCaptureTool.execute({ ...input, apiKey }, {});
33923
+ return { content: [{ type: "text", text: JSON.stringify(result ?? {}) }], structuredContent: result ?? {}, isError: false };
33924
+ });
33925
+ registerCutover("upsertTagTool", async (input, apiKey) => {
33926
+ const result = await import_upsert_tag.upsertTagTool.execute({ ...input, apiKey }, {});
33927
+ return { content: [{ type: "text", text: JSON.stringify(result ?? {}) }], structuredContent: result ?? {}, isError: false };
33928
+ });
33470
33929
  }
33471
- var import_accept_share, import_approve_sender, import_decline_share, import_inbox_settings, import_remove_approved_sender, import_revoke_share, import_set_agent_identity, import_share_note, import_share_vault, import_swap_vault, import_switch_account, import_unlink_share, import_memory_questions, import_create_channel, import_list_channel_messages, import_poll_channel, import_post_message, import_react_message, import_remove_channel_member, import_reply_message, import_record_fact, import_library_ingest, import_delete_note, import_put, import_upload, import_delete_scheduled_action, import_pause_scheduled_action, import_resume_scheduled_action, import_table_create, import_table_delete_rows, import_table_drop, import_table_insert_rows, import_add_vault, import_delete_vault, registered2;
33930
+ var import_accept_share, import_approve_sender, import_decline_share, import_inbox_settings, import_remove_approved_sender, import_revoke_share, import_set_agent_identity, import_share_note, import_share_vault, import_swap_vault, import_switch_account, import_unlink_share, import_memory_questions, import_create_channel, import_list_channel_messages, import_poll_channel, import_post_message, import_react_message, import_remove_channel_member, import_reply_message, import_record_fact, import_library_ingest, import_delete_note, import_put, import_upload, import_delete_scheduled_action, import_pause_scheduled_action, import_resume_scheduled_action, import_table_create, import_table_delete_rows, import_table_drop, import_table_insert_rows, import_add_vault, import_delete_vault, import_memory_capture, import_upsert_tag, registered2;
33472
33931
  var init_writes = __esm({
33473
33932
  "src/mcp/memory-cutover/writes.ts"() {
33474
33933
  "use strict";
@@ -33507,6 +33966,8 @@ var init_writes = __esm({
33507
33966
  import_table_insert_rows = require("mcpscraper-memory-tools/tools/tables/table-insert-rows");
33508
33967
  import_add_vault = require("mcpscraper-memory-tools/tools/vaults/add-vault");
33509
33968
  import_delete_vault = require("mcpscraper-memory-tools/tools/vaults/delete-vault");
33969
+ import_memory_capture = require("mcpscraper-memory-tools/tools/capture/memory-capture");
33970
+ import_upsert_tag = require("mcpscraper-memory-tools/tools/tags/upsert-tag");
33510
33971
  registered2 = false;
33511
33972
  }
33512
33973
  });
@@ -37811,7 +38272,7 @@ async function depositScrapeToVault(user, opts) {
37811
38272
  const clipped = content.length > MAX_DEPOSIT_CHARS ? content.slice(0, MAX_DEPOSIT_CHARS) : content;
37812
38273
  const { key, error } = await getOrCreateUserMemoryKey(user);
37813
38274
  if (!key) return { deposited: false, error: error ?? "memory unavailable" };
37814
- const vault = opts.vault?.trim() || personalVault(user);
38275
+ const vault = opts.vault?.trim() || "Library";
37815
38276
  const title = (opts.title?.trim() || opts.source).slice(0, 200);
37816
38277
  const res = await memoryCall(
37817
38278
  "libraryIngestTool",
@@ -37819,7 +38280,7 @@ async function depositScrapeToVault(user, opts) {
37819
38280
  key
37820
38281
  );
37821
38282
  if (!res.ok) return { deposited: false, vault, error: res.error ?? "ingest failed" };
37822
- return { deposited: true, vault, noteId: res.noteId, path: res.path, chunks: res.chunks };
38283
+ return { deposited: true, vault, noteId: res.noteId, path: res.path, chunks: res.indexed };
37823
38284
  } catch (err) {
37824
38285
  return { deposited: false, error: err?.message ?? "deposit failed" };
37825
38286
  }
@@ -37882,16 +38343,16 @@ async function cleanupLocal(cutoff) {
37882
38343
  let deleted = 0;
37883
38344
  let entries;
37884
38345
  try {
37885
- entries = await (0, import_promises11.readdir)(dir);
38346
+ entries = await (0, import_promises10.readdir)(dir);
37886
38347
  } catch {
37887
38348
  return { deleted: 0, store: "local" };
37888
38349
  }
37889
38350
  for (const name of entries) {
37890
38351
  const path6 = (0, import_node_path16.join)(dir, name);
37891
38352
  try {
37892
- const s = await (0, import_promises11.stat)(path6);
38353
+ const s = await (0, import_promises10.stat)(path6);
37893
38354
  if (s.isFile() && s.mtimeMs < cutoff) {
37894
- await (0, import_promises11.unlink)(path6);
38355
+ await (0, import_promises10.unlink)(path6);
37895
38356
  deleted++;
37896
38357
  }
37897
38358
  } catch {
@@ -37908,11 +38369,11 @@ async function cleanupExpiredScrapeBlobs(maxAgeMs = SCRAPE_BLOB_TTL_MS) {
37908
38369
  return { deleted: 0, store: "none" };
37909
38370
  }
37910
38371
  }
37911
- var import_promises11, import_node_os11, import_node_path16;
38372
+ var import_promises10, import_node_os11, import_node_path16;
37912
38373
  var init_scrape_blob_cleanup = __esm({
37913
38374
  "src/api/scrape-blob-cleanup.ts"() {
37914
38375
  "use strict";
37915
- import_promises11 = require("fs/promises");
38376
+ import_promises10 = require("fs/promises");
37916
38377
  import_node_os11 = require("os");
37917
38378
  import_node_path16 = require("path");
37918
38379
  init_scrape_vault_sink();
@@ -38231,7 +38692,7 @@ function countPaaQuestions(result) {
38231
38692
  return Array.isArray(value.flat) ? value.flat.length : 0;
38232
38693
  }
38233
38694
  function paaCostForQuestionCount(questionCount) {
38234
- return Math.max(1, questionCount) * MC_COSTS.paa;
38695
+ return MC_COSTS.paa_base + Math.max(1, questionCount) * MC_COSTS.paa;
38235
38696
  }
38236
38697
  async function processJob(job) {
38237
38698
  running++;