@mgsoftwarebv/mg-dashboard-mcp 7.0.23 → 7.2.0

package/dist/index.js

@@ -956,9 +956,9 @@ async function fetchAndFormatRun(conn, proxy, sshExec2, instance, runId) {
     return { content: [{ type: "text", text: `Invalid API response:
 ${rawJson.substring(0, 500)}` }] };
   }
-  let text7 = formatRunDetail(run);
-  if (logs) text7 += "\n\n--- Logs ---\n" + logs;
-  return { content: [{ type: "text", text: text7 }] };
+  let text8 = formatRunDetail(run);
+  if (logs) text8 += "\n\n--- Logs ---\n" + logs;
+  return { content: [{ type: "text", text: text8 }] };
 }
 async function waitForCompletion(conn, proxy, sshExec2, instance, runId, waitSeconds) {
   const pollInterval = 3e3;
@@ -980,10 +980,10 @@ async function waitForCompletion(conn, proxy, sshExec2, instance, runId, waitSec
       continue;
     }
     if (TERMINAL_STATUSES.has(run.status)) {
-      let text7 = formatRunDetail(run);
+      let text8 = formatRunDetail(run);
       const logs = await fetchRunLogs(runId, conn, proxy, sshExec2);
-      if (logs) text7 += "\n\n--- Logs ---\n" + logs;
-      return { content: [{ type: "text", text: text7 }] };
+      if (logs) text8 += "\n\n--- Logs ---\n" + logs;
+      return { content: [{ type: "text", text: text8 }] };
     }
   }
   return {
@@ -3670,10 +3670,10 @@ var ZodObject = class _ZodObject extends ZodType {
   //   }) as any;
   //   return merged;
   // }
-  catchall(index5) {
+  catchall(index6) {
     return new _ZodObject({
       ...this._def,
-      catchall: index5
+      catchall: index6
     });
   }
   pick(mask) {
@@ -3991,9 +3991,9 @@ function mergeValues(a, b) {
       return { valid: false };
     }
     const newArray = [];
-    for (let index5 = 0; index5 < a.length; index5++) {
-      const itemA = a[index5];
-      const itemB = b[index5];
+    for (let index6 = 0; index6 < a.length; index6++) {
+      const itemA = a[index6];
+      const itemB = b[index6];
       const sharedValue = mergeValues(itemA, itemB);
       if (!sharedValue.valid) {
         return { valid: false };
@@ -4199,10 +4199,10 @@ var ZodMap = class extends ZodType {
     }
     const keyType = this._def.keyType;
     const valueType = this._def.valueType;
-    const pairs = [...ctx.data.entries()].map(([key, value], index5) => {
+    const pairs = [...ctx.data.entries()].map(([key, value], index6) => {
       return {
-        key: keyType._parse(new ParseInputLazyPath(ctx, key, ctx.path, [index5, "key"])),
-        value: valueType._parse(new ParseInputLazyPath(ctx, value, ctx.path, [index5, "value"]))
+        key: keyType._parse(new ParseInputLazyPath(ctx, key, ctx.path, [index6, "key"])),
+        value: valueType._parse(new ParseInputLazyPath(ctx, value, ctx.path, [index6, "value"]))
       };
     });
     if (ctx.common.async) {
@@ -5170,6 +5170,222 @@ pgTable("two_factor", {
   secret: text("secret").notNull(),
   backupCodes: text("backup_codes")
 });
+var contentSourceType = pgEnum("content_source_type", [
+  "own_site",
+  "client_site",
+  "external_source",
+  "competitor",
+  "news",
+  "government",
+  "supplier",
+  "unknown"
+]);
+var contentSnapshotKind = pgEnum("content_snapshot_kind", [
+  "corpus",
+  "source"
+]);
+var contentSources = pgTable(
+  "content_sources",
+  {
+    id: uuid("id").primaryKey().defaultRandom(),
+    url: text("url").notNull(),
+    canonicalUrl: text("canonical_url"),
+    domain: text("domain").notNull(),
+    sourceType: contentSourceType("source_type").notNull().default("unknown"),
+    title: text("title"),
+    author: text("author"),
+    publishedAt: timestamp("published_at", { withTimezone: true }),
+    extractionMethod: text("extraction_method"),
+    summary: text("summary"),
+    facts: jsonb("facts").$type().notNull().default([]),
+    limitations: jsonb("limitations").$type().notNull().default([]),
+    fullText: text("full_text"),
+    textHash: text("text_hash"),
+    wordCount: integer("word_count"),
+    language: text("language"),
+    projectKey: text("project_key"),
+    siteKey: text("site_key"),
+    ticketNumber: text("ticket_number"),
+    metadata: jsonb("metadata").$type().notNull().default({}),
+    fetchedAt: timestamp("fetched_at", { withTimezone: true }).notNull().defaultNow(),
+    lastSeenAt: timestamp("last_seen_at", { withTimezone: true }).notNull().defaultNow(),
+    createdAt: timestamp("created_at", { withTimezone: true }).notNull().defaultNow(),
+    updatedAt: timestamp("updated_at", { withTimezone: true }).notNull().defaultNow()
+  },
+  (table) => [
+    uniqueIndex("content_sources_canonical_url_uidx").on(table.canonicalUrl),
+    index("content_sources_domain_idx").on(table.domain),
+    index("content_sources_type_idx").on(table.sourceType),
+    index("content_sources_text_hash_idx").on(table.textHash),
+    index("content_sources_project_idx").on(table.projectKey),
+    index("content_sources_ticket_idx").on(table.ticketNumber),
+    index("content_sources_last_seen_idx").on(table.lastSeenAt.desc())
+  ]
+);
+var contentEvidencePacks = pgTable(
+  "content_evidence_packs",
+  {
+    id: uuid("id").primaryKey().defaultRandom(),
+    topic: text("topic"),
+    locale: text("locale"),
+    projectKey: text("project_key"),
+    siteKey: text("site_key"),
+    ticketNumber: text("ticket_number"),
+    generatedAt: timestamp("generated_at", { withTimezone: true }).notNull().defaultNow(),
+    sources: jsonb("sources").$type().notNull().default([]),
+    verifiedFacts: jsonb("verified_facts").$type().notNull().default([]),
+    conflicts: jsonb("conflicts").$type().notNull().default([]),
+    uniqueClaims: jsonb("unique_claims").$type().notNull().default([]),
+    openQuestions: jsonb("open_questions").$type().notNull().default([]),
+    verbatimWarnings: jsonb("verbatim_warnings").$type().notNull().default([]),
+    limitations: jsonb("limitations").$type().notNull().default([]),
+    metadata: jsonb("metadata").$type().notNull().default({}),
+    createdAt: timestamp("created_at", { withTimezone: true }).notNull().defaultNow(),
+    updatedAt: timestamp("updated_at", { withTimezone: true }).notNull().defaultNow()
+  },
+  (table) => [
+    index("content_evidence_packs_project_idx").on(table.projectKey),
+    index("content_evidence_packs_ticket_idx").on(table.ticketNumber),
+    index("content_evidence_packs_generated_idx").on(table.generatedAt.desc())
+  ]
+);
+pgTable(
+  "content_evidence_pack_source",
+  {
+    id: uuid("id").primaryKey().defaultRandom(),
+    packId: uuid("pack_id").notNull().references(() => contentEvidencePacks.id, { onDelete: "cascade" }),
+    contentSourceId: uuid("content_source_id").notNull().references(() => contentSources.id, { onDelete: "cascade" }),
+    packSourceRef: text("pack_source_ref"),
+    createdAt: timestamp("created_at", { withTimezone: true }).notNull().defaultNow()
+  },
+  (table) => [
+    uniqueIndex("content_evidence_pack_source_uidx").on(
+      table.packId,
+      table.contentSourceId
+    ),
+    index("content_evidence_pack_source_source_idx").on(table.contentSourceId)
+  ]
+);
+var contentCorpusItems = pgTable(
+  "content_corpus_items",
+  {
+    id: uuid("id").primaryKey().defaultRandom(),
+    projectKey: text("project_key").notNull(),
+    siteKey: text("site_key"),
+    url: text("url").notNull(),
+    canonicalUrl: text("canonical_url"),
+    domain: text("domain").notNull(),
+    pagePath: text("page_path"),
+    sourceType: contentSourceType("source_type").notNull().default("own_site"),
+    title: text("title"),
+    h1: text("h1"),
+    headings: jsonb("headings").$type().notNull().default([]),
+    wordCount: integer("word_count"),
+    fullText: text("full_text"),
+    textHash: text("text_hash"),
+    internalLinks: jsonb("internal_links").$type().notNull().default([]),
+    externalLinks: jsonb("external_links").$type().notNull().default([]),
+    language: text("language"),
+    ticketNumber: text("ticket_number"),
+    metadata: jsonb("metadata").$type().notNull().default({}),
+    extractedAt: timestamp("extracted_at", { withTimezone: true }).notNull().defaultNow(),
+    createdAt: timestamp("created_at", { withTimezone: true }).notNull().defaultNow(),
+    updatedAt: timestamp("updated_at", { withTimezone: true }).notNull().defaultNow()
+  },
+  (table) => [
+    uniqueIndex("content_corpus_items_project_canonical_uidx").on(
+      table.projectKey,
+      table.canonicalUrl
+    ),
+    index("content_corpus_items_domain_idx").on(table.domain),
+    index("content_corpus_items_project_idx").on(table.projectKey),
+    index("content_corpus_items_type_idx").on(table.sourceType),
+    index("content_corpus_items_text_hash_idx").on(table.textHash),
+    index("content_corpus_items_extracted_idx").on(table.extractedAt.desc())
+  ]
+);
+pgTable(
+  "content_quality_runs",
+  {
+    id: uuid("id").primaryKey().defaultRandom(),
+    corpusItemId: uuid("corpus_item_id").references(
+      () => contentCorpusItems.id,
+      { onDelete: "set null" }
+    ),
+    contentSourceId: uuid("content_source_id").references(
+      () => contentSources.id,
+      { onDelete: "set null" }
+    ),
+    projectKey: text("project_key"),
+    siteKey: text("site_key"),
+    url: text("url"),
+    canonicalUrl: text("canonical_url"),
+    tool: text("tool").notNull(),
+    toolVersion: text("tool_version"),
+    score: integer("score"),
+    flags: jsonb("flags").$type().notNull().default([]),
+    report: jsonb("report").$type().notNull().default({}),
+    reportUrl: text("report_url"),
+    ticketNumber: text("ticket_number"),
+    runAt: timestamp("run_at", { withTimezone: true }).notNull().defaultNow(),
+    createdAt: timestamp("created_at", { withTimezone: true }).notNull().defaultNow()
+  },
+  (table) => [
+    index("content_quality_runs_corpus_idx").on(table.corpusItemId),
+    index("content_quality_runs_source_idx").on(table.contentSourceId),
+    index("content_quality_runs_project_run_idx").on(
+      table.projectKey,
+      table.runAt.desc()
+    ),
+    index("content_quality_runs_url_idx").on(table.url)
+  ]
+);
+pgTable(
+  "content_snapshots",
+  {
+    id: uuid("id").primaryKey().defaultRandom(),
+    corpusItemId: uuid("corpus_item_id").references(
+      () => contentCorpusItems.id,
+      { onDelete: "cascade" }
+    ),
+    contentSourceId: uuid("content_source_id").references(
+      () => contentSources.id,
+      { onDelete: "cascade" }
+    ),
+    kind: contentSnapshotKind("kind").notNull(),
+    url: text("url"),
+    canonicalUrl: text("canonical_url"),
+    domain: text("domain"),
+    title: text("title"),
+    fullText: text("full_text"),
+    textHash: text("text_hash"),
+    wordCount: integer("word_count"),
+    headings: jsonb("headings").$type().notNull().default([]),
+    internalLinks: jsonb("internal_links").$type().notNull().default([]),
+    metadata: jsonb("metadata").$type().notNull().default({}),
+    capturedAt: timestamp("captured_at", { withTimezone: true }).notNull().defaultNow(),
+    createdAt: timestamp("created_at", { withTimezone: true }).notNull().defaultNow()
+  },
+  (table) => [
+    index("content_snapshots_corpus_idx").on(
+      table.corpusItemId,
+      table.capturedAt.desc()
+    ),
+    index("content_snapshots_source_idx").on(
+      table.contentSourceId,
+      table.capturedAt.desc()
+    ),
+    index("content_snapshots_text_hash_idx").on(table.textHash),
+    uniqueIndex("content_snapshots_corpus_hash_uidx").on(
+      table.corpusItemId,
+      table.textHash
+    ),
+    uniqueIndex("content_snapshots_source_hash_uidx").on(
+      table.contentSourceId,
+      table.textHash
+    )
+  ]
+);
 var managedServerOs = pgEnum("managed_server_os", [
   "linux",
   "windows",
@@ -7075,7 +7291,7 @@ function getEncryptionKey() {
     throw new Error("ENCRYPTION_KEY must be a 64-character hex string");
   return buf;
 }
-function encrypt(text7) {
+function encrypt(text8) {
   const key = getEncryptionKey();
   const iv = randomBytes(ENC_IV_LENGTH);
   const cipher = createCipheriv(
@@ -7083,7 +7299,7 @@ function encrypt(text7) {
     new Uint8Array(key),
     new Uint8Array(iv)
   );
-  let encrypted = cipher.update(text7, "utf8", "hex");
+  let encrypted = cipher.update(text8, "utf8", "hex");
   encrypted += cipher.final("hex");
   const authTag = cipher.getAuthTag();
   return Buffer.concat([
@@ -7803,10 +8019,10 @@ async function r2GetObjectRange(bucket, key, range) {
     const body = result.Body;
     if (!body?.transformToString)
       throw new Error("R2 returned no readable body");
-    const text7 = await body.transformToString();
+    const text8 = await body.transformToString();
     const header = `# range: bytes ${range.offset}-${end} of ${size} (${effectiveLen} bytes)`;
     return `${header}
-${text7}`;
+${text8}`;
   } catch (e) {
     throw r2WrapError(bucket, key, e);
   }
@@ -8169,15 +8385,15 @@ async function sftpRead(opts, filePath, proxy, options) {
             clearTimeout(timer);
             cleanup?.();
             cleanup = void 0;
-            const text7 = Buffer.concat(
+            const text8 = Buffer.concat(
               chunks.map((ch) => new Uint8Array(ch))
             ).toString("utf-8");
             if (!isWholeFileRequest) {
               const header = `# range: bytes ${offset}-${offset + effectiveLen - 1} of ${total} (${effectiveLen} bytes)`;
               resolve(`${header}
-${text7}`);
+${text8}`);
             } else {
-              resolve(text7);
+              resolve(text8);
             }
           });
           rs.on("error", (e) => {
@@ -8208,7 +8424,25 @@ function formatBytes(bytes) {
 }
 var RESPONSE_MAX_BYTES = 8192;
 var NO_FOOTER_TOOLS = /* @__PURE__ */ new Set();
-var RAW_JSON_TOOLS = /* @__PURE__ */ new Set(["get_mg_dashboard_commits"]);
+var RAW_JSON_TOOLS = /* @__PURE__ */ new Set([
+  "get_mg_dashboard_commits",
+  "extract_article",
+  "research_topic",
+  // Content registry / corpus (2026-DASMG-041) — compact JSON consumed verbatim.
+  "save_content_source",
+  "get_content_source_by_url",
+  "get_content_source_by_id",
+  "search_content_sources",
+  "save_research_pack",
+  "get_research_pack",
+  "save_content_corpus_item",
+  "list_project_content_corpus",
+  "get_content_context_for_project",
+  "save_content_quality_run",
+  "list_content_snapshots",
+  "link_content_pack_to_ticket",
+  "prune_content_snapshots"
+]);
 var TOOL_CACHE_TTL_MS = {
   "list-servers": 6e4,
   "docker-list": 3e4
@@ -8308,11 +8542,11 @@ function getKnownContainers(serverId, maxAgeMs = 5 * 6e4) {
   if (Date.now() - e.capturedAt > maxAgeMs) return void 0;
   return e.names;
 }
-function truncateForLLM(text7, maxBytes) {
-  const totalBytes = Buffer.byteLength(text7, "utf8");
+function truncateForLLM(text8, maxBytes) {
+  const totalBytes = Buffer.byteLength(text8, "utf8");
   if (totalBytes <= maxBytes)
-    return { text: text7, truncated: false, totalBytes, shownBytes: totalBytes };
-  const buf = Buffer.from(text7, "utf8");
+    return { text: text8, truncated: false, totalBytes, shownBytes: totalBytes };
+  const buf = Buffer.from(text8, "utf8");
   let cut = maxBytes;
   while (cut > 0 && (buf[cut] & 192) === 128) cut--;
   const head = buf.subarray(0, cut).toString("utf8");
@@ -8343,10 +8577,10 @@ function postprocessResult(result, meta) {
   if (!result.content?.length) return result;
   if (RAW_JSON_TOOLS.has(meta.toolName)) return result;
   const block = result.content[0];
-  let text7 = String(block.text ?? "");
-  const trunc = truncateForLLM(text7, RESPONSE_MAX_BYTES);
+  let text8 = String(block.text ?? "");
+  const trunc = truncateForLLM(text8, RESPONSE_MAX_BYTES);
   if (trunc.truncated) {
-    text7 = trunc.text + "\n\n... " + buildTruncationHint(
+    text8 = trunc.text + "\n\n... " + buildTruncationHint(
       meta.toolName,
       meta.args,
       trunc.totalBytes,
@@ -8360,11 +8594,11 @@ function postprocessResult(result, meta) {
     const parts = [`took ${tookStr}`, sizeStr];
     if (meta.serverIdLabel) parts.push(`server: ${meta.serverIdLabel}`);
     if (meta.cached) parts.push("cached");
-    text7 = `${text7}
+    text8 = `${text8}
 
 [${parts.join(", ")}]`;
   }
-  return { ...result, content: [{ ...block, text: text7 }] };
+  return { ...result, content: [{ ...block, text: text8 }] };
 }
 function buildPipelineScript(commands, shell, marker, stopOnError) {
   if (shell === "powershell") {
@@ -10356,6 +10590,377 @@ var TOOLS = [
       required: ["actor", "dateFrom", "dateTo"]
     }
   },
+  {
+    name: "extract_article",
+    description: "Extract readable article/content data from ONE PUBLIC URL, with a browser-render fallback for pages that block a normal fetch (403) or only render with JavaScript (ticket 2026-DASMG-039). Returns title, author, publishedAt, source, summary, facts[], optional rawText (capped), canonicalUrl, the `extractionMethod` used (http | browser | jsonld | rss | amp | metadata_only) and an explicit `limitations` array (e.g. blocked_without_browser, paywall_detected, partial_content, rawtext_truncated). PUBLIC URLs only \u2014 SSRF-guarded (private/loopback/metadata addresses rejected), no paywall bypass, no credentials; the source/canonical URL is always preserved. Use it to turn a news/article link into safe factual source material for agents; it never auto-publishes or rewrites. For several sources + cross-checking, use research_topic.",
+    inputSchema: {
+      type: "object",
+      properties: {
+        url: {
+          type: "string",
+          description: "Public http(s) URL of the article/page to extract."
+        },
+        includeRawText: {
+          type: "boolean",
+          description: "Include the readable body as `rawText` (default true). Set false for a compact metadata-only result."
+        },
+        maxRawTextChars: {
+          type: "number",
+          description: "Hard cap for `rawText` length (default 20000, max 100000). Truncation adds a `rawtext_truncated` limitation."
+        }
+      },
+      required: ["url"]
+    }
+  },
+  {
+    name: "research_topic",
+    description: "Turn SEVERAL public sources on one topic into a single cross-checked, sourced 'evidence pack' (ticket 2026-DASMG-040). Composes extract_article per URL and aggregates: collect, extract, cross-check, then hand a downstream agent a safe base to write ORIGINAL content from. Modes: curated (`urls` + optional `topic`), hybrid (`urls` + `discover`), discovery (`topic` + `discover.enabled`). Discovery is currently DEFERRED \u2014 when requested it returns a `discovery_unavailable` limitation while curated `urls` still work. Returns `sources` (id,url,canonicalUrl,source,title,publishedAt,authorityTier,extractionMethod,limitations), `verifiedFacts` (claim + corroboratedBy + confidence + authorityTier + asOf), `conflicts` (variants + suggestedResolution), `uniqueClaims` (single-source), `openQuestions`, `verbatimWarnings`, and pack-level `limitations`. Structured EVIDENCE ONLY \u2014 never ready-to-publish prose or brand voice; every claim traces back to a source id. PUBLIC URLs only (inherits the SSRF guard). Persistence/reuse is out of scope (2026-DASMG-041).",
+    inputSchema: {
+      type: "object",
+      properties: {
+        topic: {
+          type: "string",
+          description: "The topic the evidence pack is about (used for context/labelling)."
+        },
+        urls: {
+          type: "array",
+          items: { type: "string" },
+          description: "Curated list of public http(s) URLs to extract and cross-check (max 20)."
+        },
+        discover: {
+          type: "object",
+          description: "Discovery options (currently deferred). enabled=true emits a discovery_unavailable limitation; curated urls still work.",
+          properties: {
+            enabled: { type: "boolean" },
+            maxSources: { type: "number" },
+            recencyDays: { type: "number" },
+            locale: { type: "string" },
+            allowDomains: { type: "array", items: { type: "string" } },
+            blockDomains: { type: "array", items: { type: "string" } }
+          }
+        },
+        includeRawText: {
+          type: "boolean",
+          description: "Reserved for compatibility; claim mining always reads body text up to maxRawTextChars."
+        },
+        maxRawTextChars: {
+          type: "number",
+          description: "Per-source body cap used for claim mining (default 4000, max 20000)."
+        }
+      }
+    }
+  },
+  // ----- Content registry / corpus persistence (2026-DASMG-041) -----
+  {
+    name: "save_content_source",
+    description: "Persist (dedupe + upsert) ONE fetched source \u2014 the durable companion to extract_article (2026-DASMG-041). Pass a `url` (re-extracts internally) and/or already-extracted fields (title/summary/facts/fullText). Dedupes on canonical URL then text hash, and snapshots the previous version into content_snapshots when the text changed. External sources keep their `source_type` so they stay labelled as evidence, never own content. Returns the stored id + a compact summary + whether it was created/updated/snapshotted.",
+    inputSchema: {
+      type: "object",
+      properties: {
+        url: { type: "string", description: "Source URL (required)." },
+        canonicalUrl: { type: "string", description: "Canonical URL used for dedupe." },
+        sourceType: {
+          type: "string",
+          enum: [
+            "own_site",
+            "client_site",
+            "external_source",
+            "competitor",
+            "news",
+            "government",
+            "supplier",
+            "unknown"
+          ],
+          description: "Provenance label. own_site/client_site = managed content; the rest are evidence only."
+        },
+        title: { type: "string", description: "Title." },
+        author: { type: "string", description: "Author." },
+        publishedAt: { type: "string", description: "Publish date (any parseable format)." },
+        summary: { type: "string", description: "Short factual summary." },
+        facts: { type: "array", items: { type: "string" }, description: "Key facts." },
+        limitations: { type: "array", items: { type: "string" } },
+        fullText: { type: "string", description: "Extracted body text." },
+        language: { type: "string", description: "Language code." },
+        projectKey: { type: "string", description: "Project/site identifier." },
+        siteKey: { type: "string", description: "Site identifier." },
+        ticketNumber: { type: "string", description: "Related ticket number." },
+        metadata: { type: "object" },
+        reExtract: {
+          type: "boolean",
+          description: "Force a re-fetch even when fields are supplied."
+        }
+      },
+      required: ["url"]
+    }
+  },
+  {
+    name: "get_content_source_by_url",
+    description: "Look up a stored source by URL or canonical URL (2026-DASMG-041). Compact by default (id + summary + counts); set `includeFullText` for the body. Use to answer 'was this link already scanned?' before re-extracting.",
+    inputSchema: {
+      type: "object",
+      properties: {
+        url: { type: "string", description: "URL or canonical URL." },
+        includeFullText: { type: "boolean", description: "Return full_text." }
+      },
+      required: ["url"]
+    }
+  },
+  {
+    name: "get_content_source_by_id",
+    description: "Fetch a stored source by id (2026-DASMG-041). Compact by default; set `includeFullText` for the body.",
+    inputSchema: {
+      type: "object",
+      properties: {
+        id: { type: "string", description: "content_sources.id (uuid)." },
+        includeFullText: { type: "boolean", description: "Return full_text." }
+      },
+      required: ["id"]
+    }
+  },
+  {
+    name: "search_content_sources",
+    description: "Full-text + filter search over stored sources (2026-DASMG-041): query (title/summary/body), domain, sourceType, projectKey, ticketNumber, dateFrom/dateTo. Compact, paginated results. Use to reuse evidence already in the registry.",
+    inputSchema: {
+      type: "object",
+      properties: {
+        query: { type: "string", description: "Free-text search (title/summary/body)." },
+        domain: { type: "string", description: "Exact domain filter." },
+        sourceType: {
+          type: "string",
+          enum: [
+            "own_site",
+            "client_site",
+            "external_source",
+            "competitor",
+            "news",
+            "government",
+            "supplier",
+            "unknown"
+          ]
+        },
+        projectKey: { type: "string", description: "Project filter." },
+        ticketNumber: { type: "string", description: "Ticket filter." },
+        dateFrom: { type: "string", description: "Last-seen lower bound (ISO/date)." },
+        dateTo: { type: "string", description: "Last-seen upper bound (ISO/date)." },
+        limit: { type: "integer", description: "Max rows (default 20, max 100)." },
+        offset: { type: "integer", description: "Pagination offset." },
+        includeFullText: { type: "boolean", description: "Return full_text per row." }
+      }
+    }
+  },
+  {
+    name: "save_research_pack",
+    description: "Persist a research_topic evidence pack (2026-DASMG-041): cross-checked verified facts, conflicts, unique claims, open questions. Pass an already-built `pack`, or `urls`/`topic` to research then save. Each source is deduped into content_sources and linked via the join table, so packs can be reused without re-crawling. Returns the pack id + compact counts + linked source ids.",
+    inputSchema: {
+      type: "object",
+      properties: {
+        pack: { type: "object", description: "A research_topic result payload." },
+        urls: {
+          type: "array",
+          items: { type: "string" },
+          description: "URLs to research if no pack is given."
+        },
+        topic: { type: "string", description: "Topic to research if no pack is given." },
+        locale: { type: "string", description: "Locale of the pack." },
+        projectKey: { type: "string", description: "Project identifier." },
+        siteKey: { type: "string", description: "Site identifier." },
+        ticketNumber: { type: "string", description: "Related ticket number." },
+        sourceType: {
+          type: "string",
+          enum: [
+            "own_site",
+            "client_site",
+            "external_source",
+            "competitor",
+            "news",
+            "government",
+            "supplier",
+            "unknown"
+          ]
+        },
+        metadata: { type: "object" }
+      }
+    }
+  },
+  {
+    name: "get_research_pack",
+    description: "Fetch a stored evidence pack by id (2026-DASMG-041). Compact by default (counts + provenance); set `includeSources` / `includeClaims` for the full arrays.",
+    inputSchema: {
+      type: "object",
+      properties: {
+        id: { type: "string", description: "content_evidence_packs.id (uuid)." },
+        includeSources: { type: "boolean", description: "Return the source list." },
+        includeClaims: {
+          type: "boolean",
+          description: "Return verified facts / conflicts / unique claims."
+        }
+      },
+      required: ["id"]
+    }
+  },
+  {
+    name: "save_content_corpus_item",
+    description: "Upsert ONE own/managed page into the corpus (2026-DASMG-041), keyed by (projectKey, canonical url) \u2014 the current state of content you own. Pass a `url` (re-extracts) and/or title/h1/headings/fullText/internalLinks. Snapshots the previous version when the text changed. Use this for your/your client's pages, not external evidence.",
+    inputSchema: {
+      type: "object",
+      properties: {
+        projectKey: { type: "string", description: "Project identifier (required)." },
+        url: { type: "string", description: "Page URL (required)." },
+        siteKey: { type: "string", description: "Site identifier." },
+        canonicalUrl: { type: "string", description: "Canonical URL." },
+        sourceType: {
+          type: "string",
+          enum: [
+            "own_site",
+            "client_site",
+            "external_source",
+            "competitor",
+            "news",
+            "government",
+            "supplier",
+            "unknown"
+          ],
+          description: "Defaults to own_site."
+        },
+        pagePath: { type: "string", description: "Path within the site." },
+        title: { type: "string", description: "Title." },
+        h1: { type: "string", description: "Main heading." },
+        headings: {
+          type: "array",
+          items: { type: "object" },
+          description: "Headings ({level,text})."
+        },
+        wordCount: { type: "integer", description: "Word count." },
+        fullText: { type: "string", description: "Full page text." },
+        internalLinks: {
+          type: "array",
+          items: { type: "object" },
+          description: "Internal links ({url,text})."
+        },
+        externalLinks: { type: "array", items: { type: "object" } },
+        language: { type: "string", description: "Language code." },
+        ticketNumber: { type: "string", description: "Related ticket number." },
+        metadata: { type: "object" },
+        reExtract: {
+          type: "boolean",
+          description: "Force a re-fetch even when fields are supplied."
+        }
+      },
+      required: ["projectKey", "url"]
+    }
+  },
+  {
+    name: "list_project_content_corpus",
+    description: "List corpus items for a project (2026-DASMG-041), filter by domain/sourceType/query. Compact, paginated. Use to see content coverage and reuse own pages.",
+    inputSchema: {
+      type: "object",
+      properties: {
+        projectKey: { type: "string", description: "Project identifier (required)." },
+        domain: { type: "string", description: "Domain filter." },
+        sourceType: {
+          type: "string",
+          enum: [
+            "own_site",
+            "client_site",
+            "external_source",
+            "competitor",
+            "news",
+            "government",
+            "supplier",
+            "unknown"
+          ]
+        },
+        query: { type: "string", description: "Free-text search." },
+        limit: { type: "integer", description: "Max rows (default 20)." },
+        offset: { type: "integer", description: "Pagination offset." },
+        includeFullText: { type: "boolean", description: "Return full_text per row." }
+      },
+      required: ["projectKey"]
+    }
+  },
+  {
+    name: "get_content_context_for_project",
+    description: "Aggregate context for a content agent (2026-DASMG-041): recent corpus items, recent evidence packs and related sources for a project (optionally focused by topic or url), plus totals. One call to brief an agent on what already exists before it writes.",
+    inputSchema: {
+      type: "object",
+      properties: {
+        projectKey: { type: "string", description: "Project identifier (required)." },
+        topic: { type: "string", description: "Optional topic focus." },
+        url: { type: "string", description: "Optional url focus." },
+        limit: { type: "integer", description: "Max items per section (default 10)." }
+      },
+      required: ["projectKey"]
+    }
+  },
+  {
+    name: "save_content_quality_run",
+    description: "Store a reference to an MG SEO CLI / audit run (2026-DASMG-041): score, flags and report, attached to a corpus item, source or url. MG Dashboard stores the run; the CLI does the scoring (the CLI stays the QC layer).",
+    inputSchema: {
+      type: "object",
+      properties: {
+        tool: { type: "string", description: "Tool name, e.g. mg-seo (required)." },
+        corpusItemId: { type: "string", description: "content_corpus_items.id." },
+        contentSourceId: { type: "string", description: "content_sources.id." },
+        projectKey: { type: "string", description: "Project identifier." },
+        siteKey: { type: "string", description: "Site identifier." },
+        url: { type: "string", description: "Audited URL." },
+        canonicalUrl: { type: "string", description: "Canonical URL." },
+        toolVersion: { type: "string", description: "Tool version." },
+        score: { type: "integer", description: "Overall score 0-100." },
+        flags: {
+          type: "array",
+          items: { type: "object" },
+          description: "Flags ({code,severity,message})."
+        },
+        report: { type: "object", description: "Full report payload." },
+        reportUrl: { type: "string", description: "Link to the full report." },
+        ticketNumber: { type: "string", description: "Related ticket number." },
+        runAt: { type: "string", description: "Run timestamp (ISO)." }
+      },
+      required: ["tool"]
+    }
+  },
+  {
+    name: "list_content_snapshots",
+    description: "List historical snapshots for a corpus item or source (2026-DASMG-041) by id, url or canonical url, newest first. Compact; set `includeFullText` for the captured body. Use to see how content changed over time.",
+    inputSchema: {
+      type: "object",
+      properties: {
+        corpusItemId: { type: "string", description: "content_corpus_items.id." },
+        contentSourceId: { type: "string", description: "content_sources.id." },
+        url: { type: "string", description: "URL to match." },
+        canonicalUrl: { type: "string", description: "Canonical URL to match." },
+        limit: { type: "integer", description: "Max rows (default 20)." },
+        includeFullText: { type: "boolean", description: "Return captured full_text." }
+      }
+    }
+  },
+  {
+    name: "link_content_pack_to_ticket",
+    description: "Attach a stored evidence pack to a ticket number (2026-DASMG-041).",
+    inputSchema: {
+      type: "object",
+      properties: {
+        packId: { type: "string", description: "content_evidence_packs.id (uuid)." },
+        ticketNumber: { type: "string", description: "Ticket number." }
+      },
+      required: ["packId", "ticketNumber"]
+    }
+  },
+  {
+    name: "prune_content_snapshots",
+    description: "Retention cleanup (2026-DASMG-041): keep the newest N snapshots per item and drop the rest older than the retention window. Defaults: keepPerItem 20, olderThanDays 365. Returns how many were pruned.",
+    inputSchema: {
+      type: "object",
+      properties: {
+        keepPerItem: { type: "integer", description: "Snapshots to keep per item (default 20)." },
+        olderThanDays: {
+          type: "integer",
+          description: "Only prune snapshots older than this many days (default 365)."
+        }
+      }
+    }
+  },
   {
     name: "search-team-memory",
     description: "Search the team's ENTIRE past Cursor history (every developer, every project) for how something was handled before. Use this FIRST, before investigating from scratch, whenever you: hit a non-trivial bug or error, are about to build something that may have been done before, need a project-specific convention/gotcha, or the user asks 'have we done X', 'how did we fix Y', or 'did we solve this already'. Hybrid semantic + keyword search over mirrored conversations. Returns ranked past chats with repo, date, a solution snippet and a similarity score. Phrase the query as the problem in natural language (e.g. 'release pipeline PM2 deploy fails with module not found').",
@@ -10585,6 +11190,121 @@ async function executeToolCall(name, a, _serverId) {
           content: [{ type: "text", text: JSON.stringify(data, null, 2) }]
         };
       }
+      // ----- Public-web extraction (article tools) -----
+      case "extract_article": {
+        const url = typeof a.url === "string" ? a.url.trim() : "";
+        if (!url) {
+          return { content: [{ type: "text", text: "Error: url is required" }] };
+        }
+        const res = await fetch(`${dashboardBaseUrl}/api/tools/extract-article`, {
+          method: "POST",
+          headers: {
+            "content-type": "application/json",
+            authorization: `Bearer ${apiKey}`
+          },
+          body: JSON.stringify({
+            url,
+            includeRawText: typeof a.includeRawText === "boolean" ? a.includeRawText : void 0,
+            maxRawTextChars: typeof a.maxRawTextChars === "number" ? a.maxRawTextChars : void 0
+          })
+        });
+        if (!res.ok) {
+          const detail = await res.text().catch(() => "");
+          return {
+            content: [
+              {
+                type: "text",
+                text: `Error: extract_article failed (${res.status}). ${detail.slice(0, 300)}`
+              }
+            ]
+          };
+        }
+        const data = await res.json();
+        return {
+          content: [{ type: "text", text: JSON.stringify(data, null, 2) }]
+        };
+      }
+      case "research_topic": {
+        const urls = Array.isArray(a.urls) ? a.urls.filter((u) => typeof u === "string") : void 0;
+        const topic = typeof a.topic === "string" ? a.topic.trim() : void 0;
+        if ((!urls || urls.length === 0) && !topic) {
+          return {
+            content: [
+              { type: "text", text: "Error: provide `urls` (curated) and/or a `topic`" }
+            ]
+          };
+        }
+        const res = await fetch(`${dashboardBaseUrl}/api/tools/research-topic`, {
+          method: "POST",
+          headers: {
+            "content-type": "application/json",
+            authorization: `Bearer ${apiKey}`
+          },
+          body: JSON.stringify({
+            topic,
+            urls,
+            discover: a.discover && typeof a.discover === "object" ? a.discover : void 0,
+            includeRawText: typeof a.includeRawText === "boolean" ? a.includeRawText : void 0,
+            maxRawTextChars: typeof a.maxRawTextChars === "number" ? a.maxRawTextChars : void 0
+          })
+        });
+        if (!res.ok) {
+          const detail = await res.text().catch(() => "");
+          return {
+            content: [
+              {
+                type: "text",
+                text: `Error: research_topic failed (${res.status}). ${detail.slice(0, 300)}`
+              }
+            ]
+          };
+        }
+        const data = await res.json();
+        return {
+          content: [{ type: "text", text: JSON.stringify(data, null, 2) }]
+        };
+      }
+      // ----- Content registry / corpus persistence (2026-DASMG-041) -----
+      // All content-registry tools proxy to one dispatcher route; the tool name
+      // is the dispatcher `action` and the args are forwarded verbatim as
+      // `params` (the route validates them with the matching Zod schema).
+      case "save_content_source":
+      case "get_content_source_by_url":
+      case "get_content_source_by_id":
+      case "search_content_sources":
+      case "save_research_pack":
+      case "get_research_pack":
+      case "save_content_corpus_item":
+      case "list_project_content_corpus":
+      case "get_content_context_for_project":
+      case "save_content_quality_run":
+      case "list_content_snapshots":
+      case "link_content_pack_to_ticket":
+      case "prune_content_snapshots": {
+        const res = await fetch(`${dashboardBaseUrl}/api/tools/content-registry`, {
+          method: "POST",
+          headers: {
+            "content-type": "application/json",
+            authorization: `Bearer ${apiKey}`
+          },
+          body: JSON.stringify({ action: name, params: a })
+        });
+        if (!res.ok) {
+          const detail = await res.text().catch(() => "");
+          return {
+            content: [
+              {
+                type: "text",
+                text: `Error: ${name} failed (${res.status}). ${detail.slice(0, 300)}`
+              }
+            ]
+          };
+        }
+        const data = await res.json();
+        return {
+          content: [{ type: "text", text: JSON.stringify(data, null, 2) }]
+        };
+      }
       // ----- Team memory -----
       case "search-team-memory": {
         const query = typeof a.query === "string" ? a.query.trim() : "";
@@ -10626,7 +11346,7 @@ async function executeToolCall(name, a, _serverId) {
             ]
           };
         }
-        const lines = data.hits.map((hit, index5) => {
+        const lines = data.hits.map((hit, index6) => {
           const when = hit.lastMessageAt ? new Date(hit.lastMessageAt).toISOString().slice(0, 10) : "unknown date";
           const repo = hit.repo ?? "unknown repo";
           const sim = hit.similarity !== null ? ` \xB7 ${(hit.similarity * 100).toFixed(0)}% match` : "";
@@ -10636,7 +11356,7 @@ async function executeToolCall(name, a, _serverId) {
             ...Array.isArray(hit.tech) ? hit.tech.slice(0, 4) : []
           ].filter(Boolean);
           const tags = facets.length > 0 ? ` \xB7 ${facets.join(", ")}` : "";
-          return `${index5 + 1}. [${repo} \xB7 ${when}${sim}${tags}] ${hit.title}
+          return `${index6 + 1}. [${repo} \xB7 ${when}${sim}${tags}] ${hit.title}
    id: ${hit.id}
    ${(hit.summary ?? hit.snippet).replace(/\s+/g, " ").slice(0, 500)}`;
         });
@@ -11821,8 +12541,8 @@ ${sample.join("\n")}${files.length > 5 ? `
             };
           const filtered = sortRows(applyFilter(only.rows));
           if (format === "json") {
-            const text8 = filtered.map((c) => JSON.stringify(c)).join("\n") || "(no containers)";
-            return { content: [{ type: "text", text: text8 }] };
+            const text9 = filtered.map((c) => JSON.stringify(c)).join("\n") || "(no containers)";
+            return { content: [{ type: "text", text: text9 }] };
           }
           if (groupByProject) {
             const groups = /* @__PURE__ */ new Map();
@@ -11849,8 +12569,8 @@ ${sample.join("\n")}${files.length > 5 ? `
           }
           const header = `${"NAMES".padEnd(36)}  ${"IMAGE".padEnd(40)}  ${"STATUS".padEnd(24)}  ${"HEALTH".padEnd(10)}  ${"PORTS".padEnd(40)}  PROJECT`;
           const body = filtered.map((r) => `${fmtRow(r)}  ${r.Project || ""}`);
-          const text7 = filtered.length === 0 ? "(no containers match)" : [header, ...body].join("\n");
-          return { content: [{ type: "text", text: text7 }] };
+          const text8 = filtered.length === 0 ? "(no containers match)" : [header, ...body].join("\n");
+          return { content: [{ type: "text", text: text8 }] };
         }
         if (format === "json") {
           const lines = [];