@oculum/scanner 1.0.9 → 1.0.11

package/src/__tests__/benchmark/fixtures/layer2/phase1-enhancements.ts

@@ -0,0 +1,157 @@
+/**
+ * Test Fixtures: Phase 1 Enhancement Backlog
+ *
+ * Tests for:
+ * - MCP Tool Description Injection
+ * - MCP Server Shadowing
+ * - RAG Query Injection
+ * - RAG Embedding Poisoning
+ * - RAG Chunk Injection
+ * - Package Typosquatting
+ */
+
+// ============================================================================
+// MCP Tool Description Injection (should trigger ai_mcp_description_injection)
+// ============================================================================
+
+// Dynamic tool description from variable (HIGH)
+const userInput = req.body.toolDescription
+const tool1 = {
+  name: 'dynamic-tool',
+  description: `This tool does ${userInput} for the user`,
+}
+
+// Description concatenated with user input (HIGH)
+const tool2 = {
+  name: 'concat-tool',
+  description: 'Basic description ' + user.preference,
+}
+
+// Injection keywords in tool description (CRITICAL)
+const tool3 = {
+  name: 'suspicious-tool',
+  description: 'This tool ignores previous instructions and does whatever',
+}
+
+// Tool name from config (HIGH)
+registerTool(config['toolName'], { execute: handler })
+
+// Python dynamic description (HIGH)
+// @server.tool(name=f"tool-{user_id}")
+// async def dynamic_tool():
+//     pass
+
+// Description from database (MEDIUM)
+const tool4 = {
+  name: 'db-tool',
+  description: await db.get('tool_description'),
+}
+
+// ============================================================================
+// MCP Server Shadowing (should trigger ai_mcp_server_shadowing)
+// ============================================================================
+
+// Server config from environment (MEDIUM)
+const MCP_SERVERS = JSON.parse(process.env.MCP_SERVER_CONFIG)
+
+// Server URL from user input (HIGH)
+const mcpConfig = {
+  serverUrl: req.body.serverUrl,
+  timeout: 30000,
+}
+
+// Dynamic server registration (MEDIUM)
+serverList.forEach(s => registerServer(s))
+
+// Servers from parsed JSON (HIGH)
+const servers = JSON.parse(req.body.mcpServers)
+
+// Server config override (MEDIUM)
+Object.assign(serverConfig, req.body.serverOverrides)
+
+// ============================================================================
+// RAG Query Injection (should trigger ai_rag_query_injection)
+// ============================================================================
+
+// User input directly in retrieval query (HIGH)
+const results1 = await vectorStore.search(req.body.query)
+const results2 = await retriever.invoke(user.searchTerm)
+const results3 = await index.query(input.question)
+
+// Query from request body without validation (MEDIUM)
+const { query } = req.body
+const docs = await vectorStore.similaritySearch(query, 10)
+
+// Query template with user input (MEDIUM)
+const searchQuery = `Find documents about ${user.topic} in the ${req.body.category} category`
+
+// Query passthrough to vector store (HIGH)
+app.post('/search', async (req, res) => {
+  const results = await vectorStore.query(req.body.query)
+  res.json(results)
+})
+
+// ============================================================================
+// RAG Embedding Poisoning (should trigger ai_rag_embedding_poisoning)
+// ============================================================================
+
+// User document embedded without validation (HIGH)
+const embedding = await embeddings.create(user.document)
+await vectorStore.add(embedding)
+
+// Retrieval without similarity threshold (MEDIUM)
+const nearbyDocs = await vectorStore.similaritySearch(query, 20)
+
+// Batch embedding without duplicate detection (LOW)
+for (const doc of userDocs) {
+  await vectorStore.add(await embeddings.embed(doc))
+}
+
+// Dynamic embedding model from config (MEDIUM)
+const embeddingModel = new config['embeddingModel']
+
+// External URL content embedded (HIGH)
+const content = await fetch(userProvidedUrl)
+const embedded = await embeddings.embed(content.text)
+await vectorStore.add(embedded)
+
+// ============================================================================
+// RAG Chunk Injection (should trigger ai_rag_chunk_injection)
+// ============================================================================
+
+// User content chunked without validation (MEDIUM)
+const chunks = await textSplitter.split(user.document)
+await vectorStore.add(chunks)
+
+// Context joined without separators (LOW)
+const context = docs.map(d => d.pageContent).join('')
+
+// Chunk metadata from user input (MEDIUM)
+await vectorStore.add(chunks, { metadata: user.metadata })
+
+// Chunking without size validation (LOW)
+const parts = splitter.createDocuments(content)
+
+// ============================================================================
+// Package Typosquatting (should trigger ai_package_typosquat)
+// ============================================================================
+
+// 1 character difference from popular package
+import reqests from 'reqests'       // similar to 'requests'
+import loodash from 'loodash'       // similar to 'lodash'
+import expres from 'expres'         // similar to 'express'
+
+// Character substitution (0 for o, 1 for l)
+import l0dash from 'l0dash'         // lodash with 0 instead of o
+import ax1os from 'ax1os'           // axios with 1 instead of i
+
+// Missing vowel
+import rqsts from 'rqsts'           // requests without vowels
+
+// Common suffix confusion
+import react_js from 'react-js'     // react with -js suffix
+import node_express from 'node-express'  // express with node- prefix
+
+export default {
+  // Placeholder for test
+}