visus-mcp 0.2.0 → 0.6.0

package/dist/utils/format-converter.js

@@ -0,0 +1,191 @@
+/**
+ * Format Converter - Content-Type based format detection and conversion
+ *
+ * Handles format-appropriate conversion based on detected Content-Type.
+ * Supports HTML, JSON, XML, and RSS/Atom feeds.
+ */
+import { XMLParser } from 'fast-xml-parser';
+/**
+ * Detect format from Content-Type header
+ *
+ * @param contentType - Content-Type header value (e.g., "application/json", "text/html; charset=utf-8")
+ * @returns Detected format type
+ */
+export function detectFormat(contentType) {
+    // Normalize: lowercase and extract MIME type (before semicolon)
+    const mimeType = contentType.toLowerCase().split(';')[0].trim();
+    // HTML formats
+    if (mimeType === 'text/html' || mimeType === 'application/xhtml+xml') {
+        return 'html';
+    }
+    // JSON formats
+    if (mimeType === 'application/json' || mimeType === 'text/json') {
+        return 'json';
+    }
+    // RSS/Atom feed formats
+    if (mimeType === 'application/rss+xml' ||
+        mimeType === 'application/atom+xml' ||
+        mimeType === 'application/feed+json') {
+        return 'rss';
+    }
+    // XML formats (must come after RSS check)
+    if (mimeType === 'application/xml' || mimeType === 'text/xml') {
+        return 'xml';
+    }
+    // Default to HTML for unknown types
+    return 'html';
+}
+/**
+ * Convert JSON content to formatted string
+ *
+ * @param raw - Raw JSON string
+ * @returns Formatted JSON string with prefix, or raw string if parse fails
+ */
+export function convertJson(raw) {
+    try {
+        // Parse and re-stringify with 2-space indent for readability
+        const parsed = JSON.parse(raw);
+        const formatted = JSON.stringify(parsed, null, 2);
+        return `JSON Response:\n\n${formatted}`;
+    }
+    catch (error) {
+        // Parse failed, return raw string unchanged
+        return raw;
+    }
+}
+/**
+ * Convert XML content to clean text representation
+ *
+ * @param raw - Raw XML string
+ * @returns Formatted XML representation with prefix, or tag-stripped fallback if parse fails
+ */
+export function convertXml(raw) {
+    try {
+        const parser = new XMLParser({
+            ignoreAttributes: false,
+            attributeNamePrefix: '@_',
+            textNodeName: '#text',
+            ignoreDeclaration: true,
+            ignorePiTags: true,
+            removeNSPrefix: true,
+        });
+        const parsed = parser.parse(raw);
+        const formatted = JSON.stringify(parsed, null, 2);
+        return `XML Response:\n\n${formatted}`;
+    }
+    catch (error) {
+        // Parse failed, strip XML tags using regex and return
+        const stripped = raw.replace(/<[^>]+>/g, '').trim();
+        return `XML Response:\n\n${stripped}`;
+    }
+}
+/**
+ * Convert RSS/Atom feed content to clean Markdown
+ *
+ * @param raw - Raw RSS/Atom XML string
+ * @returns Formatted Markdown representation, or falls back to convertXml if parse fails
+ */
+export function convertRss(raw) {
+    try {
+        const parser = new XMLParser({
+            ignoreAttributes: false,
+            attributeNamePrefix: '@_',
+            textNodeName: '#text',
+            removeNSPrefix: true,
+        });
+        const parsed = parser.parse(raw);
+        // Handle RSS 2.0 format
+        if (parsed.rss && parsed.rss.channel) {
+            return formatRss2(parsed.rss.channel);
+        }
+        // Handle Atom format
+        if (parsed.feed) {
+            return formatAtom(parsed.feed);
+        }
+        // Handle RSS 1.0 (RDF) format
+        if (parsed.rdf && parsed.rdf.channel) {
+            return formatRss2(parsed.rdf.channel);
+        }
+        // Unknown feed format, fall back to XML
+        return convertXml(raw);
+    }
+    catch (error) {
+        // Parse failed, fall back to XML converter
+        return convertXml(raw);
+    }
+}
+/**
+ * Format RSS 2.0 feed data as Markdown
+ */
+function formatRss2(channel) {
+    const title = channel.title || 'Untitled Feed';
+    const description = channel.description || '';
+    const items = Array.isArray(channel.item) ? channel.item : (channel.item ? [channel.item] : []);
+    let markdown = `RSS Feed:\n\n# ${title}\n`;
+    if (description) {
+        markdown += `${description}\n`;
+    }
+    markdown += '\n## Items\n\n';
+    // Extract up to 10 items
+    const itemsToShow = items.slice(0, 10);
+    for (const item of itemsToShow) {
+        const itemTitle = item.title || 'Untitled';
+        const itemLink = item.link || '';
+        const itemDescription = item.description || '';
+        const itemPubDate = item.pubDate || '';
+        // Truncate description to 200 chars
+        const truncatedDesc = itemDescription.length > 200
+            ? itemDescription.substring(0, 200) + '...'
+            : itemDescription;
+        markdown += `### ${itemTitle}\n\n`;
+        if (truncatedDesc) {
+            markdown += `${truncatedDesc}\n\n`;
+        }
+        if (itemLink) {
+            markdown += `Link: ${itemLink}\n`;
+        }
+        if (itemPubDate) {
+            markdown += `Published: ${itemPubDate}\n`;
+        }
+        markdown += '\n---\n\n';
+    }
+    return markdown;
+}
+/**
+ * Format Atom feed data as Markdown
+ */
+function formatAtom(feed) {
+    const title = feed.title || 'Untitled Feed';
+    const subtitle = feed.subtitle || '';
+    const entries = Array.isArray(feed.entry) ? feed.entry : (feed.entry ? [feed.entry] : []);
+    let markdown = `RSS Feed:\n\n# ${title}\n`;
+    if (subtitle) {
+        markdown += `${subtitle}\n`;
+    }
+    markdown += '\n## Items\n\n';
+    // Extract up to 10 entries
+    const entriesToShow = entries.slice(0, 10);
+    for (const entry of entriesToShow) {
+        const entryTitle = entry.title || 'Untitled';
+        const entryLink = entry.link ? (entry.link['@_href'] || entry.link) : '';
+        const entrySummary = entry.summary || entry.content || '';
+        const entryPublished = entry.published || entry.updated || '';
+        // Truncate summary to 200 chars
+        const truncatedSummary = entrySummary.length > 200
+            ? entrySummary.substring(0, 200) + '...'
+            : entrySummary;
+        markdown += `### ${entryTitle}\n\n`;
+        if (truncatedSummary) {
+            markdown += `${truncatedSummary}\n\n`;
+        }
+        if (entryLink) {
+            markdown += `Link: ${entryLink}\n`;
+        }
+        if (entryPublished) {
+            markdown += `Published: ${entryPublished}\n`;
+        }
+        markdown += '\n---\n\n';
+    }
+    return markdown;
+}
+//# sourceMappingURL=format-converter.js.map

package/dist/utils/format-converter.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"format-converter.js","sourceRoot":"","sources":["../../src/utils/format-converter.ts"],"names":[],"mappings":"AAAA;;;;;GAKG;AAEH,OAAO,EAAE,SAAS,EAAE,MAAM,iBAAiB,CAAC;AAO5C;;;;;GAKG;AACH,MAAM,UAAU,YAAY,CAAC,WAAmB;IAC9C,gEAAgE;IAChE,MAAM,QAAQ,GAAG,WAAW,CAAC,WAAW,EAAE,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC;IAEhE,eAAe;IACf,IAAI,QAAQ,KAAK,WAAW,IAAI,QAAQ,KAAK,uBAAuB,EAAE,CAAC;QACrE,OAAO,MAAM,CAAC;IAChB,CAAC;IAED,eAAe;IACf,IAAI,QAAQ,KAAK,kBAAkB,IAAI,QAAQ,KAAK,WAAW,EAAE,CAAC;QAChE,OAAO,MAAM,CAAC;IAChB,CAAC;IAED,wBAAwB;IACxB,IAAI,QAAQ,KAAK,qBAAqB;QAClC,QAAQ,KAAK,sBAAsB;QACnC,QAAQ,KAAK,uBAAuB,EAAE,CAAC;QACzC,OAAO,KAAK,CAAC;IACf,CAAC;IAED,0CAA0C;IAC1C,IAAI,QAAQ,KAAK,iBAAiB,IAAI,QAAQ,KAAK,UAAU,EAAE,CAAC;QAC9D,OAAO,KAAK,CAAC;IACf,CAAC;IAED,oCAAoC;IACpC,OAAO,MAAM,CAAC;AAChB,CAAC;AAED;;;;;GAKG;AACH,MAAM,UAAU,WAAW,CAAC,GAAW;IACrC,IAAI,CAAC;QACH,6DAA6D;QAC7D,MAAM,MAAM,GAAG,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC;QAC/B,MAAM,SAAS,GAAG,IAAI,CAAC,SAAS,CAAC,MAAM,EAAE,IAAI,EAAE,CAAC,CAAC,CAAC;QAClD,OAAO,qBAAqB,SAAS,EAAE,CAAC;IAC1C,CAAC;IAAC,OAAO,KAAK,EAAE,CAAC;QACf,4CAA4C;QAC5C,OAAO,GAAG,CAAC;IACb,CAAC;AACH,CAAC;AAED;;;;;GAKG;AACH,MAAM,UAAU,UAAU,CAAC,GAAW;IACpC,IAAI,CAAC;QACH,MAAM,MAAM,GAAG,IAAI,SAAS,CAAC;YAC3B,gBAAgB,EAAE,KAAK;YACvB,mBAAmB,EAAE,IAAI;YACzB,YAAY,EAAE,OAAO;YACrB,iBAAiB,EAAE,IAAI;YACvB,YAAY,EAAE,IAAI;YAClB,cAAc,EAAE,IAAI;SACrB,CAAC,CAAC;QAEH,MAAM,MAAM,GAAG,MAAM,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC;QACjC,MAAM,SAAS,GAAG,IAAI,CAAC,SAAS,CAAC,MAAM,EAAE,IAAI,EAAE,CAAC,CAAC,CAAC;QAElD,OAAO,oBAAoB,SAAS,EAAE,CAAC;IACzC,CAAC;IAAC,OAAO,KAAK,EAAE,CAAC;QACf,sDAAsD;QACtD,MAAM,QAAQ,GAAG,GAAG,CAAC,OAAO,CAAC,UAAU,EAAE,EAAE,CAAC,CAAC,IAAI,EAAE,CAAC;QACpD,OAAO,oBAAoB,QAAQ,EAAE,CAAC;IACxC,CAAC;AACH,CAAC;AAED;;;;;GAKG;AACH,MAAM,UAAU,UAAU,CAAC,GAAW;IACpC,IAAI,CAAC;QACH,MAAM,MAAM,GAAG,IAAI,SAAS,CAAC;YAC3B,gBAAgB,EAAE,KAAK;YACvB,mBAAmB,EAAE,IAAI;YACzB,YAAY,EAAE,OAAO;YACrB,cAAc,EAAE,IAAI;SACrB,CAAC,CAAC;QAEH,MAAM,MAAM,GAAG,MAAM,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC;QAEjC,wBAAwB;QACxB,IAAI,MAAM,CAAC,GAAG,IAAI,MAAM,CAAC,GAAG,CAAC,OAAO,EAAE,CAAC;YACrC,OAAO,UAAU,CAAC,MAAM,CAAC,GAAG,CAAC,OAAO,CAAC,CAAC;QACxC,CAAC;QAED,qBAAqB;QACrB,IAAI,MAAM,CAAC,IAAI,EAAE,CAAC;YAChB,OAAO,UAAU,CAAC,MAAM,CAAC,IAAI,CAAC,CAAC;QACjC,CAAC;QAED,8BAA8B;QAC9B,IAAI,MAAM,CAAC,GAAG,IAAI,MAAM,CAAC,GAAG,CAAC,OAAO,EAAE,CAAC;YACrC,OAAO,UAAU,CAAC,MAAM,CAAC,GAAG,CAAC,OAAO,CAAC,CAAC;QACxC,CAAC;QAED,wCAAwC;QACxC,OAAO,UAAU,CAAC,GAAG,CAAC,CAAC;IAEzB,CAAC;IAAC,OAAO,KAAK,EAAE,CAAC;QACf,2CAA2C;QAC3C,OAAO,UAAU,CAAC,GAAG,CAAC,CAAC;IACzB,CAAC;AACH,CAAC;AAED;;GAEG;AACH,SAAS,UAAU,CAAC,OAAY;IAC9B,MAAM,KAAK,GAAG,OAAO,CAAC,KAAK,IAAI,eAAe,CAAC;IAC/C,MAAM,WAAW,GAAG,OAAO,CAAC,WAAW,IAAI,EAAE,CAAC;IAC9C,MAAM,KAAK,GAAG,KAAK,CAAC,OAAO,CAAC,OAAO,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC;IAEhG,IAAI,QAAQ,GAAG,kBAAkB,KAAK,IAAI,CAAC;IAE3C,IAAI,WAAW,EAAE,CAAC;QAChB,QAAQ,IAAI,GAAG,WAAW,IAAI,CAAC;IACjC,CAAC;IAED,QAAQ,IAAI,gBAAgB,CAAC;IAE7B,yBAAyB;IACzB,MAAM,WAAW,GAAG,KAAK,CAAC,KAAK,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC;IAEvC,KAAK,MAAM,IAAI,IAAI,WAAW,EAAE,CAAC;QAC/B,MAAM,SAAS,GAAG,IAAI,CAAC,KAAK,IAAI,UAAU,CAAC;QAC3C,MAAM,QAAQ,GAAG,IAAI,CAAC,IAAI,IAAI,EAAE,CAAC;QACjC,MAAM,eAAe,GAAG,IAAI,CAAC,WAAW,IAAI,EAAE,CAAC;QAC/C,MAAM,WAAW,GAAG,IAAI,CAAC,OAAO,IAAI,EAAE,CAAC;QAEvC,oCAAoC;QACpC,MAAM,aAAa,GAAG,eAAe,CAAC,MAAM,GAAG,GAAG;YAChD,CAAC,CAAC,eAAe,CAAC,SAAS,CAAC,CAAC,EAAE,GAAG,CAAC,GAAG,KAAK;YAC3C,CAAC,CAAC,eAAe,CAAC;QAEpB,QAAQ,IAAI,OAAO,SAAS,MAAM,CAAC;QAEnC,IAAI,aAAa,EAAE,CAAC;YAClB,QAAQ,IAAI,GAAG,aAAa,MAAM,CAAC;QACrC,CAAC;QAED,IAAI,QAAQ,EAAE,CAAC;YACb,QAAQ,IAAI,SAAS,QAAQ,IAAI,CAAC;QACpC,CAAC;QAED,IAAI,WAAW,EAAE,CAAC;YAChB,QAAQ,IAAI,cAAc,WAAW,IAAI,CAAC;QAC5C,CAAC;QAED,QAAQ,IAAI,WAAW,CAAC;IAC1B,CAAC;IAED,OAAO,QAAQ,CAAC;AAClB,CAAC;AAED;;GAEG;AACH,SAAS,UAAU,CAAC,IAAS;IAC3B,MAAM,KAAK,GAAG,IAAI,CAAC,KAAK,IAAI,eAAe,CAAC;IAC5C,MAAM,QAAQ,GAAG,IAAI,CAAC,QAAQ,IAAI,EAAE,CAAC;IACrC,MAAM,OAAO,GAAG,KAAK,CAAC,OAAO,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC;IAE1F,IAAI,QAAQ,GAAG,kBAAkB,KAAK,IAAI,CAAC;IAE3C,IAAI,QAAQ,EAAE,CAAC;QACb,QAAQ,IAAI,GAAG,QAAQ,IAAI,CAAC;IAC9B,CAAC;IAED,QAAQ,IAAI,gBAAgB,CAAC;IAE7B,2BAA2B;IAC3B,MAAM,aAAa,GAAG,OAAO,CAAC,KAAK,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC;IAE3C,KAAK,MAAM,KAAK,IAAI,aAAa,EAAE,CAAC;QAClC,MAAM,UAAU,GAAG,KAAK,CAAC,KAAK,IAAI,UAAU,CAAC;QAC7C,MAAM,SAAS,GAAG,KAAK,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,KAAK,CAAC,IAAI,CAAC,QAAQ,CAAC,IAAI,KAAK,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC;QACzE,MAAM,YAAY,GAAG,KAAK,CAAC,OAAO,IAAI,KAAK,CAAC,OAAO,IAAI,EAAE,CAAC;QAC1D,MAAM,cAAc,GAAG,KAAK,CAAC,SAAS,IAAI,KAAK,CAAC,OAAO,IAAI,EAAE,CAAC;QAE9D,gCAAgC;QAChC,MAAM,gBAAgB,GAAG,YAAY,CAAC,MAAM,GAAG,GAAG;YAChD,CAAC,CAAC,YAAY,CAAC,SAAS,CAAC,CAAC,EAAE,GAAG,CAAC,GAAG,KAAK;YACxC,CAAC,CAAC,YAAY,CAAC;QAEjB,QAAQ,IAAI,OAAO,UAAU,MAAM,CAAC;QAEpC,IAAI,gBAAgB,EAAE,CAAC;YACrB,QAAQ,IAAI,GAAG,gBAAgB,MAAM,CAAC;QACxC,CAAC;QAED,IAAI,SAAS,EAAE,CAAC;YACd,QAAQ,IAAI,SAAS,SAAS,IAAI,CAAC;QACrC,CAAC;QAED,IAAI,cAAc,EAAE,CAAC;YACnB,QAAQ,IAAI,cAAc,cAAc,IAAI,CAAC;QAC/C,CAAC;QAED,QAAQ,IAAI,WAAW,CAAC;IAC1B,CAAC;IAED,OAAO,QAAQ,CAAC;AAClB,CAAC"}

package/dist/utils/truncate.d.ts

@@ -0,0 +1,26 @@
+/**
+ * Token-aware content truncation utility
+ *
+ * Anthropic MCP Directory enforces a 25,000 token response limit.
+ * This utility provides safe truncation with token estimation.
+ */
+/**
+ * Truncate content if it exceeds the token ceiling
+ *
+ * @param content Content to potentially truncate
+ * @returns Truncated content and metadata
+ */
+export declare function truncateContent(content: string): {
+    content: string;
+    truncated: boolean;
+    truncated_at_chars?: number;
+};
+/**
+ * Estimate token count for a given string
+ * Uses conservative 4 chars per token approximation
+ *
+ * @param text Text to estimate
+ * @returns Estimated token count
+ */
+export declare function estimateTokens(text: string): number;
+//# sourceMappingURL=truncate.d.ts.map

package/dist/utils/truncate.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"truncate.d.ts","sourceRoot":"","sources":["../../src/utils/truncate.ts"],"names":[],"mappings":"AAAA;;;;;GAKG;AAmBH;;;;;GAKG;AACH,wBAAgB,eAAe,CAAC,OAAO,EAAE,MAAM,GAAG;IAChD,OAAO,EAAE,MAAM,CAAC;IAChB,SAAS,EAAE,OAAO,CAAC;IACnB,kBAAkB,CAAC,EAAE,MAAM,CAAC;CAC7B,CAkBA;AAED;;;;;;GAMG;AACH,wBAAgB,cAAc,CAAC,IAAI,EAAE,MAAM,GAAG,MAAM,CAEnD"}

package/dist/utils/truncate.js

@@ -0,0 +1,54 @@
+/**
+ * Token-aware content truncation utility
+ *
+ * Anthropic MCP Directory enforces a 25,000 token response limit.
+ * This utility provides safe truncation with token estimation.
+ */
+/**
+ * Maximum tokens allowed in MCP response (Anthropic Directory limit)
+ * We target 24,000 to leave headroom for metadata/JSON structure
+ */
+const MAX_TOKENS = 24000;
+/**
+ * Conservative token estimation: 1 token ≈ 4 characters
+ * This is a safe approximation that errs on the side of caution
+ */
+const CHARS_PER_TOKEN = 4;
+/**
+ * Maximum characters based on token limit
+ */
+const MAX_CHARS = MAX_TOKENS * CHARS_PER_TOKEN; // 96,000 characters
+/**
+ * Truncate content if it exceeds the token ceiling
+ *
+ * @param content Content to potentially truncate
+ * @returns Truncated content and metadata
+ */
+export function truncateContent(content) {
+    if (content.length <= MAX_CHARS) {
+        // Content is within limits
+        return {
+            content,
+            truncated: false
+        };
+    }
+    // Content exceeds limit - truncate with warning message
+    const truncatedContent = content.substring(0, MAX_CHARS);
+    const warningMessage = `\n\n--- CONTENT TRUNCATED ---\nOriginal length: ${content.length} characters (~${Math.ceil(content.length / CHARS_PER_TOKEN)} tokens)\nTruncated to: ${MAX_CHARS} characters (~${MAX_TOKENS} tokens)\nReason: Anthropic MCP Directory enforces a 25,000 token response limit\n`;
+    return {
+        content: truncatedContent + warningMessage,
+        truncated: true,
+        truncated_at_chars: MAX_CHARS
+    };
+}
+/**
+ * Estimate token count for a given string
+ * Uses conservative 4 chars per token approximation
+ *
+ * @param text Text to estimate
+ * @returns Estimated token count
+ */
+export function estimateTokens(text) {
+    return Math.ceil(text.length / CHARS_PER_TOKEN);
+}
+//# sourceMappingURL=truncate.js.map

package/dist/utils/truncate.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"truncate.js","sourceRoot":"","sources":["../../src/utils/truncate.ts"],"names":[],"mappings":"AAAA;;;;;GAKG;AAEH;;;GAGG;AACH,MAAM,UAAU,GAAG,KAAK,CAAC;AAEzB;;;GAGG;AACH,MAAM,eAAe,GAAG,CAAC,CAAC;AAE1B;;GAEG;AACH,MAAM,SAAS,GAAG,UAAU,GAAG,eAAe,CAAC,CAAC,oBAAoB;AAEpE;;;;;GAKG;AACH,MAAM,UAAU,eAAe,CAAC,OAAe;IAK7C,IAAI,OAAO,CAAC,MAAM,IAAI,SAAS,EAAE,CAAC;QAChC,2BAA2B;QAC3B,OAAO;YACL,OAAO;YACP,SAAS,EAAE,KAAK;SACjB,CAAC;IACJ,CAAC;IAED,wDAAwD;IACxD,MAAM,gBAAgB,GAAG,OAAO,CAAC,SAAS,CAAC,CAAC,EAAE,SAAS,CAAC,CAAC;IACzD,MAAM,cAAc,GAAG,mDAAmD,OAAO,CAAC,MAAM,iBAAiB,IAAI,CAAC,IAAI,CAAC,OAAO,CAAC,MAAM,GAAG,eAAe,CAAC,2BAA2B,SAAS,iBAAiB,UAAU,oFAAoF,CAAC;IAExS,OAAO;QACL,OAAO,EAAE,gBAAgB,GAAG,cAAc;QAC1C,SAAS,EAAE,IAAI;QACf,kBAAkB,EAAE,SAAS;KAC9B,CAAC;AACJ,CAAC;AAED;;;;;;GAMG;AACH,MAAM,UAAU,cAAc,CAAC,IAAY;IACzC,OAAO,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC,MAAM,GAAG,eAAe,CAAC,CAAC;AAClD,CAAC"}

package/infrastructure/stack.ts

@@ -19,6 +19,7 @@
 
 import * as cdk from 'aws-cdk-lib';
 import * as lambda from 'aws-cdk-lib/aws-lambda';
+import * as lambdaNodejs from 'aws-cdk-lib/aws-lambda-nodejs';
 import * as apigateway from 'aws-cdk-lib/aws-apigateway';
 import * as cognito from 'aws-cdk-lib/aws-cognito';
 import * as dynamodb from 'aws-cdk-lib/aws-dynamodb';
@@ -69,6 +70,7 @@ export class VisusStack extends cdk.Stack {
       removalPolicy: environment === 'prod'
         ? cdk.RemovalPolicy.RETAIN
         : cdk.RemovalPolicy.DESTROY,
+      timeToLiveAttribute: 'ttl', // Auto-delete audit logs after 30 days
     });
 
     // Global Secondary Index for querying by request_id
@@ -142,12 +144,12 @@ export class VisusStack extends cdk.Stack {
     // Grant KMS decrypt access (for reading encrypted DynamoDB data if needed)
     kmsKey.grantEncryptDecrypt(lambdaRole);
 
-    // Lambda function
-    const visusFn = new lambda.Function(this, 'VisusFunction', {
+    // Lambda function (NodejsFunction with automatic bundling)
+    const visusFn = new lambdaNodejs.NodejsFunction(this, 'VisusFunction', {
       functionName: `visus-mcp-${environment}`,
       runtime: lambda.Runtime.NODEJS_20_X,
-      handler: 'index.handler',
-      code: lambda.Code.fromAsset('dist'), // Build output directory (relative to project root)
+      entry: 'src/lambda-handler.ts', // Entry point for bundler
+      handler: 'handler', // Export name in the entry file
       timeout: cdk.Duration.seconds(30), // Playwright page loads can take time
       memorySize: 1024, // Chromium requires significant memory
       reservedConcurrentExecutions: environment === 'prod' ? 100 : 10, // RULE 7: Cost protection
@@ -155,12 +157,21 @@ export class VisusStack extends cdk.Stack {
       environment: {
         AUDIT_TABLE_NAME: auditTable.tableName,
         ENVIRONMENT: environment,
+        ALLOWED_ORIGINS: 'https://claude.ai,https://app.claude.ai,http://localhost:3000',
         NODE_OPTIONS: '--enable-source-maps', // For debugging
       },
       logRetention: environment === 'prod'
         ? logs.RetentionDays.ONE_MONTH
         : logs.RetentionDays.ONE_WEEK,
       description: `Visus MCP sanitization service (${environment})`,
+      bundling: {
+        minify: false, // Keep readable for debugging
+        sourceMap: true,
+        externalModules: [
+          'playwright-core', // Playwright is huge, will be added via layer
+          '@sparticuz/chromium', // Chromium binary
+        ],
+      },
     });
 
     // ========================================
@@ -178,12 +189,44 @@ export class VisusStack extends cdk.Stack {
         metricsEnabled: true,
       },
       defaultCorsPreflightOptions: {
-        allowOrigins: apigateway.Cors.ALL_ORIGINS, // Phase 2: Open. Phase 3: Restrict to Lateos
-        allowMethods: apigateway.Cors.ALL_METHODS,
+        allowOrigins: [
+          'https://claude.ai',
+          'https://app.claude.ai',
+          'http://localhost:3000',  // local dev only
+        ],
+        allowMethods: ['POST', 'OPTIONS'],
         allowHeaders: ['Content-Type', 'Authorization'],
       },
     });
 
+    // Usage plan with rate limiting and quota
+    const usagePlan = api.addUsagePlan('VisusUsagePlan', {
+      name: `visus-usage-plan-${environment}`,
+      description: 'Rate limiting and quota management for Visus API',
+      throttle: {
+        rateLimit: 10,      // 10 requests per second
+        burstLimit: 20,     // 20 request burst
+      },
+      quota: {
+        limit: 1000,        // 1000 requests per day
+        period: apigateway.Period.DAY,
+      },
+    });
+
+    // Add deployment stage to usage plan
+    usagePlan.addApiStage({
+      stage: api.deploymentStage,
+    });
+
+    // Create API key for the usage plan
+    const apiKey = api.addApiKey('VisusApiKey', {
+      apiKeyName: `visus-api-key-${environment}`,
+      description: `API key for Visus ${environment} environment`,
+    });
+
+    // Associate API key with usage plan
+    usagePlan.addApiKey(apiKey);
+
     // Cognito authorizer
     const authorizer = new apigateway.CognitoUserPoolsAuthorizer(this, 'VisusAuthorizer', {
       cognitoUserPools: [userPool],
@@ -244,5 +287,11 @@ export class VisusStack extends cdk.Stack {
       description: 'Lambda function ARN',
       exportName: `visus-lambda-arn-${environment}`,
     });
+
+    new cdk.CfnOutput(this, 'ApiKeyId', {
+      value: apiKey.keyId,
+      description: 'API Gateway API Key ID (use aws apigateway get-api-key to retrieve value)',
+      exportName: `visus-api-key-id-${environment}`,
+    });
   }
 }

package/jest.config.js

@@ -15,6 +15,9 @@ export default {
   },
   testMatch: ['**/tests/**/*.test.ts'],
   testPathIgnorePatterns: ['/node_modules/', '/dist/'],
+  transformIgnorePatterns: [
+    'node_modules/(?!(@mozilla/readability|jsdom|@exodus/bytes|html-encoding-sniffer|@toon-format)/)',
+  ],
   testTimeout: 15000,
   forceExit: true,
   detectOpenHandles: false,

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "visus-mcp",
-  "version": "0.2.0",
-  "description": "Secure web access for Claude — sanitizes all web content before it reaches your LLM",
+  "version": "0.6.0",
+  "description": "Secure, sanitized web access for Claude. Detects prompt injection (43 patterns), redacts PII, renders JavaScript pages, generates NIST AI 600-1 / OWASP / MITRE ATLAS threat reports, and auto-detects JSON/XML/RSS content types.",
   "main": "dist/index.js",
   "bin": {
     "visus-mcp": "dist/index.js"
@@ -39,15 +39,22 @@
     "node": ">=18"
   },
   "dependencies": {
+    "@aws-sdk/client-dynamodb": "^3.1014.0",
+    "@aws-sdk/lib-dynamodb": "^3.1014.0",
     "@modelcontextprotocol/sdk": "^1.0.4",
+    "@mozilla/readability": "^0.6.0",
     "@playwright/test": "^1.58.2",
+    "@toon-format/toon": "^2.1.0",
     "cheerio": "^1.2.0",
+    "fast-xml-parser": "^5.5.8",
+    "jsdom": "^29.0.1",
     "playwright": "^1.58.2",
     "undici": "^7.24.5"
   },
   "devDependencies": {
     "@types/aws-lambda": "^8.10.161",
     "@types/jest": "^29.5.14",
+    "@types/jsdom": "^28.0.1",
     "@types/node": "^20.19.37",
     "aws-cdk": "^2.1112.0",
     "aws-cdk-lib": "^2.244.0",

package/src/browser/playwright-renderer.ts

@@ -126,6 +126,7 @@ async function renderWithLambda(
       html: body.html,
       title: body.title,
       url,
+      contentType: 'text/html', // Lambda renderer defaults to HTML
       text: undefined, // Lambda renderer doesn't extract text
     });
 
@@ -164,6 +165,12 @@ async function renderWithFetch(
 
     const html = await response.body.text();
 
+    // Capture Content-Type header
+    const contentTypeHeader = response.headers['content-type'];
+    const contentType = typeof contentTypeHeader === 'string'
+      ? contentTypeHeader.split(';')[0].trim()  // Remove charset and other params
+      : 'text/html'; // Default to HTML if missing
+
     // Extract title using regex (simple fallback)
     const titleMatch = html.match(/<title[^>]*>(.*?)<\/title>/i);
     const title = titleMatch ? titleMatch[1].trim() : '';
@@ -172,6 +179,7 @@ async function renderWithFetch(
       html,
       title,
       url,
+      contentType,
       text: undefined,
     });
 

package/src/browser/reader.ts

@@ -0,0 +1,129 @@
+/**
+ * Reader Mode - Mozilla Readability Integration
+ *
+ * Extracts clean article content from web pages using Mozilla's Readability.js.
+ * This module strips navigation, ads, and boilerplate to return main article content.
+ *
+ * CRITICAL: Content extraction happens BEFORE sanitization. The pipeline is:
+ * Playwright renders → Readability extracts → Sanitizer runs → Token ceiling applied
+ */
+
+import { Readability } from '@mozilla/readability';
+import { JSDOM } from 'jsdom';
+import type { Result } from '../types.js';
+import { Ok, Err } from '../types.js';
+
+/**
+ * Result from reader mode extraction
+ */
+export interface ReaderResult {
+  title: string;
+  byline: string | null;           // Author
+  publishedTime: string | null;     // ISO timestamp or null
+  content: string;                  // Extracted text content
+  excerpt: string | null;           // Short summary
+  wordCount: number;                // Estimated word count
+  readerModeAvailable: boolean;     // True if Readability succeeded
+}
+
+/**
+ * Extract clean article content using Mozilla Readability
+ *
+ * @param html - Rendered HTML from Playwright
+ * @param url - Original URL (required for relative link resolution)
+ * @returns Result containing extracted article or fallback to full HTML
+ */
+export function extractArticle(
+  html: string,
+  url: string
+): Result<ReaderResult, Error> {
+  try {
+    // Parse HTML with jsdom
+    const dom = new JSDOM(html, { url });
+    const document = dom.window.document;
+
+    // Attempt extraction with Readability
+    const reader = new Readability(document);
+    const article = reader.parse();
+
+    // If Readability succeeds, return extracted content
+    if (article && article.textContent) {
+      const wordCount = estimateWordCount(article.textContent);
+
+      return Ok({
+        title: article.title || 'Untitled',
+        byline: article.byline || null,
+        publishedTime: article.publishedTime || null,
+        content: article.textContent,
+        excerpt: article.excerpt || null,
+        wordCount,
+        readerModeAvailable: true
+      });
+    }
+
+    // Readability failed - fallback to raw text extraction
+    const fallbackText = extractFallbackText(document);
+    const wordCount = estimateWordCount(fallbackText);
+
+    // Extract title from <title> tag as fallback
+    const titleElement = document.querySelector('title');
+    const fallbackTitle = titleElement?.textContent?.trim() || 'Untitled';
+
+    return Ok({
+      title: fallbackTitle,
+      byline: null,
+      publishedTime: null,
+      content: fallbackText,
+      excerpt: null,
+      wordCount,
+      readerModeAvailable: false
+    });
+
+  } catch (error) {
+    return Err(
+      error instanceof Error
+        ? error
+        : new Error(`Reader extraction failed: ${String(error)}`)
+    );
+  }
+}
+
+/**
+ * Estimate word count from text content
+ *
+ * @param text - Text content to count
+ * @returns Estimated word count
+ */
+function estimateWordCount(text: string): number {
+  if (!text || text.trim().length === 0) {
+    return 0;
+  }
+
+  // Split on whitespace and filter out empty strings
+  const words = text.trim().split(/\s+/).filter(word => word.length > 0);
+  return words.length;
+}
+
+/**
+ * Fallback text extraction when Readability fails
+ *
+ * Extracts visible text from the page, skipping script/style elements.
+ *
+ * @param document - JSDOM document
+ * @returns Extracted text content
+ */
+function extractFallbackText(document: Document): string {
+  // Remove script and style elements
+  const scripts = document.querySelectorAll('script, style, noscript');
+  scripts.forEach(el => el.remove());
+
+  // Extract body text
+  const bodyText = document.body?.textContent || '';
+
+  // Clean up whitespace
+  return bodyText
+    .split('\n')
+    .map(line => line.trim())
+    .filter(line => line.length > 0)
+    .join('\n');
+}