@swarmvaultai/engine 0.6.6 → 0.6.7

package/dist/index.js

@@ -64,6 +64,13 @@ function buildManagedBlock(target) {
     ""
   ].join("\n");
 }
+function buildCursorRule() {
+  const frontmatter = YAML.stringify({
+    description: "SwarmVault graph-first repository instructions.",
+    alwaysApply: true
+  }).trimEnd();
+  return ["---", frontmatter, "---", "", buildManagedBlock("cursor").trimEnd(), ""].join("\n");
+}
 function supportsAgentHook(agent) {
   return agent === "claude" || agent === "opencode" || agent === "gemini" || agent === "copilot";
 }
@@ -690,8 +697,7 @@ async function installAgent(rootDir, agent, options = {}) {
       await upsertManagedBlock(target, buildManagedBlock("gemini"));
       break;
     case "cursor":
-      await writeOwnedFile(target, `${buildManagedBlock("cursor")}
-`);
+      await writeOwnedFile(target, buildCursorRule());
       break;
     case "aider":
       await upsertManagedBlock(target, buildManagedBlock("aider"));
@@ -1343,10 +1349,13 @@ function buildBenchmarkArtifact(input) {
   const corpusTokens = Math.max(1, Math.round(input.corpusWords * (100 / 75)));
   const perQuestion = input.perQuestion.filter((entry) => entry.queryTokens > 0).map((entry) => ({
     ...entry,
-    reduction: Number(Math.max(0, 1 - entry.queryTokens / Math.max(1, corpusTokens)).toFixed(3))
+    // Honest reduction: negative values mean graph context is larger than the
+    // full corpus, which is the truth on very small vaults. Clamping to zero
+    // hid that signal.
+    reduction: Number((1 - entry.queryTokens / Math.max(1, corpusTokens)).toFixed(3))
   }));
   const avgQueryTokens = perQuestion.length ? Math.max(1, Math.round(perQuestion.reduce((total, entry) => total + entry.queryTokens, 0) / perQuestion.length)) : 0;
-  const reductionRatio = avgQueryTokens ? Number(Math.max(0, 1 - avgQueryTokens / Math.max(1, corpusTokens)).toFixed(3)) : 0;
+  const reductionRatio = avgQueryTokens ? Number((1 - avgQueryTokens / Math.max(1, corpusTokens)).toFixed(3)) : 0;
   const uniqueVisitedNodes = new Set(perQuestion.flatMap((entry) => entry.visitedNodeIds)).size;
   const summary = {
     questionCount: input.questions.length,
@@ -1728,6 +1737,7 @@ import { pathToFileURL } from "url";
 import { Readability } from "@mozilla/readability";
 import matter3 from "gray-matter";
 import ignore from "ignore";
+import { isText } from "istextorbinary";
 import { JSDOM as JSDOM2 } from "jsdom";
 import mime from "mime-types";
 import TurndownService2 from "turndown";
@@ -2447,14 +2457,18 @@ function rubyStringContent(node) {
   const contentNode = node.descendantsOfType(["string_content", "simple_symbol", "bare_string"]).find((item) => item !== null) ?? null;
   return contentNode?.text.trim() || void 0;
 }
+function normalizePowerShellDotSourceSpecifier(raw) {
+  const unquoted = raw.replace(/^['"]+|['"]+$/g, "").trim();
+  return unquoted.replace(/^\$PSScriptRoot(?:[\\/]+|$)/i, "./");
+}
 function parsePowerShellImport(commandNode) {
   const commandName = commandNode.descendantsOfType(["command_name", "command_name_expr"]).find((item) => item !== null)?.text.trim();
   const genericTokens = commandNode.descendantsOfType("generic_token").filter((item) => item !== null).map((item) => item.text.trim());
   if (commandNode.namedChildren.some((child) => child?.type === "command_invokation_operator")) {
-    const specifier = commandName?.trim();
-    if (specifier) {
+    const raw = commandName?.trim();
+    if (raw) {
       return {
-        specifier,
+        specifier: normalizePowerShellDotSourceSpecifier(raw),
         importedSymbols: [],
         isExternal: false,
         reExport: false
@@ -2935,6 +2949,21 @@ function rustCodeAnalysis(manifest, rootNode, diagnostics) {
       imports.push(parseRustUse(child.text));
       continue;
     }
+    if (child.type === "mod_item") {
+      const hasInlineBody = child.namedChildren.some((item) => item?.type === "declaration_list");
+      if (!hasInlineBody) {
+        const modName = extractIdentifier(child.childForFieldName("name") ?? findNamedChild(child, "identifier"));
+        if (modName) {
+          imports.push({
+            specifier: `self::${modName}`,
+            importedSymbols: [],
+            isExternal: false,
+            reExport: false
+          });
+        }
+      }
+      continue;
+    }
     const name = child.type === "function_item" ? extractIdentifier(child.childForFieldName("name")) : extractIdentifier(child.childForFieldName("name"));
     if (child.type === "impl_item") {
       const traitName = normalizeSymbolReference(nodeText(child.childForFieldName("trait")));
@@ -3518,8 +3547,9 @@ function rubyCodeAnalysis(manifest, rootNode, diagnostics) {
         if (callee === "require" || callee === "require_relative") {
           const specifier = rubyStringContent(child.childForFieldName("arguments") ?? child.namedChildren.at(1) ?? null);
           if (specifier) {
+            const normalizedSpecifier = callee === "require_relative" && !specifier.startsWith(".") && !specifier.startsWith("/") ? `./${specifier}` : specifier;
             imports.push({
-              specifier,
+              specifier: normalizedSpecifier,
               importedSymbols: [],
               isExternal: callee === "require" && !specifier.startsWith("."),
               reExport: false
@@ -3766,7 +3796,7 @@ async function analyzeTreeSitterCode(manifest, content, language) {
     };
   }
   try {
-    const diagnostics = diagnosticsFromTree(tree.rootNode);
+    const diagnostics = language === "lua" ? [] : diagnosticsFromTree(tree.rootNode);
     const rationales = extractTreeSitterRationales(manifest, language, tree.rootNode);
     switch (language) {
       case "bash":
@@ -4509,7 +4539,13 @@ function modulePageTitle(manifest) {
 }
 function importResolutionCandidates(basePath, specifier, extensions) {
   const resolved = path6.posix.normalize(path6.posix.join(path6.posix.dirname(basePath), specifier));
-  if (path6.posix.extname(resolved)) {
+  const resolvedExt = path6.posix.extname(resolved);
+  if (resolvedExt) {
+    if (extensions.includes(resolvedExt)) {
+      const resolvedBase = resolved.slice(0, -resolvedExt.length);
+      const candidates = [resolved, ...extensions.map((extension) => `${resolvedBase}${extension}`)];
+      return uniqueBy(candidates, (candidate) => candidate);
+    }
     return [resolved];
   }
   const direct = extensions.map((extension) => path6.posix.normalize(`${resolved}${extension}`));
@@ -4604,7 +4640,9 @@ async function readNearestGoModulePath(startPath, cache) {
 }
 function rustModuleAlias(repoRelativePath) {
   const withoutExt = stripCodeExtension2(normalizeAlias(repoRelativePath));
-  const trimmed = withoutExt.replace(/^src\//, "").replace(/\/mod$/i, "");
+  const srcIdx = withoutExt.lastIndexOf("/src/");
+  const withinCrate = srcIdx >= 0 ? withoutExt.slice(srcIdx + "/src/".length) : withoutExt.replace(/^src\//, "");
+  const trimmed = withinCrate.replace(/\/mod$/i, "");
   if (!trimmed || trimmed === "lib" || trimmed === "main") {
     return "crate";
   }
@@ -4819,7 +4857,7 @@ function resolveRustAliases(manifest, specifier) {
   if (!currentAlias) {
     return [];
   }
-  const currentParts = currentAlias.replace(/^crate::?/, "").split("::").filter(Boolean);
+  const currentParts = currentAlias.replace(/^crate(?:::)?/, "").split("::").filter(Boolean);
   if (specifier.startsWith("self::")) {
     return [`crate${currentParts.length ? `::${currentParts.join("::")}` : ""}::${specifier.slice("self::".length)}`];
   }
@@ -6610,7 +6648,7 @@ function inferKind(mimeType, filePath, detectionOptions = {}) {
   if (mimeType === "text/csv" || mimeType === "text/tab-separated-values" || filePath.toLowerCase().endsWith(".csv") || filePath.toLowerCase().endsWith(".tsv")) {
     return "csv";
   }
-  if (mimeType.startsWith("text/") || isStructuredTextMime(mimeType) || isKnownTextPath(filePath)) {
+  if (mimeType.startsWith("text/") || isStructuredTextMime(mimeType)) {
     return "text";
   }
   if (mimeType === "application/vnd.openxmlformats-officedocument.spreadsheetml.sheet" || filePath.toLowerCase().endsWith(".xlsx")) {
@@ -6696,125 +6734,30 @@ function guessMimeType(target) {
   if (extension === ".ts" || extension === ".tsx" || extension === ".mts" || extension === ".cts") {
     return "text/typescript";
   }
-  const looked = mime.lookup(target);
-  if (looked) {
-    return looked;
-  }
-  if (isKnownTextPath(target)) {
-    return "text/plain";
-  }
-  return "application/octet-stream";
-}
-var KNOWN_TEXT_DOTFILE_NAMES = /* @__PURE__ */ new Set([
-  ".gitignore",
-  ".gitattributes",
-  ".gitkeep",
-  ".gitmodules",
-  ".editorconfig",
-  ".npmrc",
-  ".yarnrc",
-  ".prettierignore",
-  ".prettierrc",
-  ".dockerignore",
-  ".eslintignore",
-  ".eslintrc",
-  ".nvmrc",
-  ".node-version",
-  ".python-version",
-  ".ruby-version",
-  ".tool-versions"
-]);
-var KNOWN_TEXT_BASENAMES = /* @__PURE__ */ new Set([
-  "readme",
-  "license",
-  "licence",
-  "copying",
-  "unlicense",
-  "notice",
-  "authors",
-  "contributors",
-  "patents",
-  "maintainers",
-  "owners",
-  "codeowners",
-  "changelog",
-  "changes",
-  "history",
-  "news",
-  "todo",
-  "install",
-  "dockerfile",
-  "containerfile",
-  "makefile",
-  "gnumakefile",
-  "rakefile",
-  "gemfile",
-  "procfile",
-  "jenkinsfile",
-  "vagrantfile",
-  "brewfile",
-  "go.mod",
-  "go.sum",
-  "go.work",
-  "go.work.sum",
-  "cargo.lock",
-  "pipfile",
-  "pipfile.lock",
-  "poetry.lock",
-  "uv.lock",
-  "py.typed",
-  "package-lock.json",
-  "yarn.lock",
-  "pnpm-lock.yaml",
-  "composer.lock",
-  "requirements.txt"
-]);
-var KNOWN_TEXT_BASENAME_PREFIXES = ["license", "licence", "copying", "unlicense", "readme", "changelog", "dockerfile", "containerfile"];
-var KNOWN_TEXT_EXTENSIONS = /* @__PURE__ */ new Set([
-  ".toml",
-  ".lock",
-  ".tmpl",
-  ".template",
-  ".mustache",
-  ".hbs",
-  ".handlebars",
-  ".ejs",
-  ".njk",
-  ".liquid",
-  ".vim",
-  ".typed",
-  ".env",
-  ".properties",
-  ".ini",
-  ".cfg",
-  ".conf",
-  ".config",
-  ".bazel",
-  ".bzl",
-  ".bat",
-  ".cmd"
-]);
-function isKnownTextPath(target) {
-  const basename = path12.basename(target).toLowerCase();
-  if (KNOWN_TEXT_DOTFILE_NAMES.has(basename)) {
-    return true;
-  }
-  if (basename === ".env" || basename.startsWith(".env.")) {
-    return true;
-  }
-  if (KNOWN_TEXT_BASENAMES.has(basename)) {
-    return true;
+  return mime.lookup(target) || "application/octet-stream";
+}
+function refineBinaryKindWithBytes(absolutePath, currentKind, bytes) {
+  if (currentKind !== "binary") {
+    return currentKind;
   }
-  for (const prefix of KNOWN_TEXT_BASENAME_PREFIXES) {
-    if (basename === prefix || basename.startsWith(`${prefix}-`) || basename.startsWith(`${prefix}.`)) {
-      return true;
-    }
+  const sniffSlice = bytes.length > 4096 ? bytes.subarray(0, 4096) : bytes;
+  return isText(absolutePath, sniffSlice) ? "text" : currentKind;
+}
+async function refineBinaryKindWithContentSniff(absolutePath, currentKind) {
+  if (currentKind !== "binary") {
+    return currentKind;
   }
-  const extension = path12.extname(target).toLowerCase();
-  if (extension && KNOWN_TEXT_EXTENSIONS.has(extension)) {
-    return true;
+  let handle;
+  try {
+    handle = await fs11.open(absolutePath, "r");
+    const chunk = Buffer.alloc(4096);
+    const { bytesRead } = await handle.read(chunk, 0, chunk.length, 0);
+    return refineBinaryKindWithBytes(absolutePath, currentKind, chunk.subarray(0, bytesRead));
+  } catch {
+    return currentKind;
+  } finally {
+    await handle?.close().catch(() => void 0);
   }
-  return false;
 }
 function sourceGroupIdFor(prepared) {
   const originKey = prepared.originType === "url" ? prepared.url ?? prepared.title : prepared.originalPath ?? prepared.title;
@@ -7666,6 +7609,7 @@ async function collectDirectoryFiles(rootDir, inputDir, repoRoot, options) {
           sourceKind = "chat_export";
         }
       }
+      sourceKind = await refineBinaryKindWithContentSniff(absolutePath, sourceKind);
       const sourceClass = sourceClassForRelativePath(relativePath, options);
       if (!supportedDirectoryKind(sourceKind)) {
         skipped.push({ path: toPosix(path12.relative(rootDir, absolutePath)), reason: `unsupported_kind:${sourceKind}` });
@@ -8287,7 +8231,7 @@ async function prepareFileInputs(rootDir, absoluteInput, repoRoot, sourceClass)
   }
   const mimeType = guessMimeType(absoluteInput);
   const detectionOptions = await localCodeDetectionOptions(absoluteInput, payloadBytes);
-  const sourceKind = inferKind(mimeType, absoluteInput, detectionOptions);
+  const sourceKind = refineBinaryKindWithBytes(absoluteInput, inferKind(mimeType, absoluteInput, detectionOptions), payloadBytes);
   const language = inferCodeLanguage(absoluteInput, mimeType, detectionOptions);
   const storedExtension = path12.extname(absoluteInput) || `.${mime.extension(mimeType) || "bin"}`;
   let title;
@@ -9071,6 +9015,7 @@ async function importInbox(rootDir, inputDir) {
         sourceKind = "chat_export";
       }
     }
+    sourceKind = await refineBinaryKindWithContentSniff(absolutePath, sourceKind);
     if (!isSupportedInboxKind(sourceKind)) {
       skipped.push({ path: toPosix(path12.relative(rootDir, absolutePath)), reason: `unsupported_kind:${sourceKind}` });
       continue;
@@ -9268,6 +9213,7 @@ import { z as z7 } from "zod";
 
 // src/analysis.ts
 import path14 from "path";
+import nlp from "compromise";
 import { fromMarkdown } from "mdast-util-from-markdown";
 import { z as z2 } from "zod";
 var ANALYSIS_FORMAT_VERSION = 7;
@@ -9345,11 +9291,31 @@ function extractTopTerms(text, count) {
   return [...frequency.entries()].sort((left, right) => right[1] - left[1] || left[0].localeCompare(right[0])).slice(0, count).map(([token]) => token);
 }
 function extractEntities(text, count) {
-  const matches = text.match(/\b[A-Z][A-Za-z0-9-]+(?:\s+[A-Z][A-Za-z0-9-]+){0,2}\b/g) ?? [];
-  return uniqueBy(
-    matches.map((value) => normalizeWhitespace(value)),
-    (value) => value.toLowerCase()
-  ).slice(0, count);
+  const candidates = [];
+  try {
+    const doc = nlp(text);
+    const segments = [
+      doc.match("#ProperNoun+").out("array"),
+      doc.people().out("array"),
+      doc.places().out("array"),
+      doc.organizations().out("array"),
+      doc.topics().out("array")
+    ];
+    for (const segment of segments) {
+      for (const term of segment) {
+        const normalized = normalizeWhitespace(term);
+        if (normalized) {
+          candidates.push(normalized);
+        }
+      }
+    }
+  } catch {
+  }
+  if (candidates.length === 0) {
+    const matches = text.match(/\b[A-Z][A-Za-z0-9-]+(?:\s+[A-Z][A-Za-z0-9-]+){0,2}\b/g) ?? [];
+    candidates.push(...matches.map((value) => normalizeWhitespace(value)));
+  }
+  return uniqueBy(candidates, (value) => value.toLowerCase()).slice(0, count);
 }
 function detectPolarity(text) {
   if (/\b(no|not|never|cannot|can't|won't|without)\b/i.test(text)) {
@@ -11068,15 +11034,37 @@ function graphAdjacency(graph) {
   }
   return adjacency;
 }
+var NODE_TYPE_PRIORITY = {
+  concept: 6,
+  entity: 5,
+  source: 4,
+  module: 3,
+  symbol: 2,
+  rationale: 1
+};
+function nodeTypePriority(type) {
+  return NODE_TYPE_PRIORITY[type] ?? 0;
+}
+function compareLabelCandidates(left, right) {
+  const priorityDelta = nodeTypePriority(right.type) - nodeTypePriority(left.type);
+  if (priorityDelta !== 0) {
+    return priorityDelta;
+  }
+  const degreeDelta = (right.degree ?? 0) - (left.degree ?? 0);
+  if (degreeDelta !== 0) {
+    return degreeDelta;
+  }
+  return left.id.localeCompare(right.id);
+}
 function resolveNode(graph, target) {
   const normalized = normalizeTarget(target);
   const byId = nodeById(graph);
   if (byId.has(target)) {
     return byId.get(target);
   }
-  const exact = graph.nodes.find((node) => normalizeTarget(node.label) === normalized || normalizeTarget(node.id) === normalized);
-  if (exact) {
-    return exact;
+  const labelMatches = graph.nodes.filter((node) => normalizeTarget(node.label) === normalized || normalizeTarget(node.id) === normalized);
+  if (labelMatches.length) {
+    return labelMatches.sort(compareLabelCandidates)[0];
   }
   const pages = graph.pages.map((page) => ({
     page,
@@ -11085,7 +11073,7 @@ function resolveNode(graph, target) {
   if (pages[0]) {
     return primaryNodeForPage(graph, pages[0].page);
   }
-  return graph.nodes.map((node) => ({ node, score: Math.max(scoreMatch(target, node.label), scoreMatch(target, node.id)) })).filter((item) => item.score > 0).sort((left, right) => right.score - left.score || left.node.label.localeCompare(right.node.label))[0]?.node;
+  return graph.nodes.map((node) => ({ node, score: Math.max(scoreMatch(target, node.label), scoreMatch(target, node.id)) })).filter((item) => item.score > 0).sort((left, right) => right.score - left.score || compareLabelCandidates(left.node, right.node))[0]?.node;
 }
 function communityLabel(graph, communityId) {
   if (!communityId) {
@@ -17860,6 +17848,10 @@ function extractClaimSectionLines(content) {
   }
   return found ? claimLines : null;
 }
+function isClaimPlaceholderBullet(line) {
+  const trimmed = line.trim();
+  return /^-\s+No\s+claims\s+extracted\.?$/i.test(trimmed);
+}
 function structuralLintFindings(_rootDir, paths, graph, schemas, manifests, sourceProjects) {
   const manifestMap = new Map(manifests.map((manifest) => [manifest.sourceId, manifest]));
   const pageMap2 = new Map(graph.pages.map((page) => [page.id, page]));
@@ -17907,7 +17899,9 @@ function structuralLintFindings(_rootDir, paths, graph, schemas, manifests, sour
         const content = await fs19.readFile(absolutePath, "utf8");
         const claimLines = extractClaimSectionLines(content);
         if (claimLines !== null) {
-          const uncited = claimLines.filter((line) => line.startsWith("- ") && !line.includes("[source:"));
+          const uncited = claimLines.filter(
+            (line) => line.startsWith("- ") && !line.includes("[source:") && !isClaimPlaceholderBullet(line)
+          );
           if (uncited.length) {
             findings.push({
               severity: "warning",
@@ -18022,7 +18016,7 @@ async function bootstrapDemo(rootDir, input) {
 }
 
 // src/mcp.ts
-var SERVER_VERSION = "0.6.6";
+var SERVER_VERSION = "0.6.7";
 async function createMcpServer(rootDir) {
   const server = new McpServer({
     name: "swarmvault",
@@ -21115,14 +21109,9 @@ async function startGraphServer(rootDir, port, options = {}) {
           response.end(JSON.stringify({ error: "Missing explain target." }));
           return;
         }
-        try {
-          const result = await explainGraphVault(rootDir, target2);
-          response.writeHead(200, { "content-type": "application/json" });
-          response.end(JSON.stringify(result));
-        } catch (error) {
-          response.writeHead(404, { "content-type": "application/json" });
-          response.end(JSON.stringify({ error: error instanceof Error ? error.message : `Could not resolve graph target: ${target2}` }));
-        }
+        const result = await explainGraphVault(rootDir, target2);
+        response.writeHead(200, { "content-type": "application/json" });
+        response.end(JSON.stringify(result));
         return;
       }
       if (url.pathname === "/api/search") {
@@ -21261,7 +21250,8 @@ async function startGraphServer(rootDir, port, options = {}) {
       const message = error instanceof Error ? error.message : String(error);
       console.error(`[viewer] ${request.method ?? "GET"} ${url.pathname} failed: ${message}`);
       if (!response.headersSent) {
-        response.writeHead(500, { "content-type": "application/json" });
+        const status = /not found|could not resolve|cannot resolve/i.test(message) ? 404 : 500;
+        response.writeHead(status, { "content-type": "application/json" });
         response.end(JSON.stringify({ error: message }));
       } else {
         response.end();

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@swarmvaultai/engine",
-  "version": "0.6.6",
+  "version": "0.6.7",
   "description": "Core engine for SwarmVault: ingest, compile, query, lint, and provider abstractions.",
   "type": "module",
   "main": "dist/index.js",
@@ -44,10 +44,12 @@
     "@mozilla/readability": "^0.6.0",
     "@vscode/tree-sitter-wasm": "^0.3.1",
     "chokidar": "^4.0.3",
+    "compromise": "^14.14.4",
     "csv-parse": "^6.2.1",
     "fflate": "^0.8.2",
     "gray-matter": "^4.0.3",
     "ignore": "^7.0.5",
+    "istextorbinary": "^9.5.0",
     "jsdom": "^27.0.0",
     "mailparser": "^3.9.8",
     "mammoth": "^1.12.0",