Every feature designed to help you understand and navigate code faster—without compromising privacy.
Ask questions in natural language. adam understands what you mean, not just what you type.
Search for "error handling" and find try/catch blocks, error boundaries, exception handlers—even if they don't contain those words.
Code is understood as meaningful units—functions, classes, modules—not just text. Results always have complete context.
Combines multiple retrieval strategies for the best results. Find exact matches when you need them, conceptual matches when you don't.
$ adam search "how do we validate user input?"
Found 8 relevant results in 0.2s
src/validation/schemas.ts:12
│ export const userSchema = z.object({
│ email: z.string().email(),
│ password: z.string().min(8)
│ });
src/api/middleware/validate.ts:5
│ export function validateRequest(schema) {
│ return (req, res, next) => {
│ const result = schema.safeParse(req.body);
│ ...
│ }
│ }
src/components/forms/RegisterForm.tsx:23
│ // Client-side validation with react-hook-form
│ const { register, handleSubmit, errors } = useForm({
│ resolver: zodResolver(userSchema)
│ });
Simulated output. NLP queries work offline and return ranked code locations with metadata. The narrative formatting shown here represents the LLM-enhanced experience.
Understand how your code connects. Every function, class, and module mapped into a queryable graph.
Imports, exports, function calls, class inheritance—all relationships are indexed and searchable.
Search results include related code. Find the function and its callers, its dependencies, and downstream effects.
Understand the blast radius of changes. See what depends on a function before you modify it.
adam integrates directly with AI assistants through the MCP protocol. Your AI tools can search, index, and analyze your code—and adam provides the rich context they need for accurate, grounded answers.
Just add adam as an MCP server. Claude can search, index, and analyze your code without leaving the conversation.
Search results include full semantic context—summaries, relationships, and documentation. Perfect for AI assistants.
Search, index, analyze, explore symbols—all available as MCP tools your AI can use.
User: How does authentication work in this project?
Claude: Let me search the codebase for you.
Using adam.search...
Based on the code, authentication works like this:
1. JWT-based - Tokens issued at login
→ src/auth/jwt.ts:23
2. Middleware validation - Every protected route
→ src/middleware/auth.ts:45
3. Refresh tokens - 7-day rotation
→ src/auth/refresh.ts:12
Would you like me to explain any part in detail?
Simulated output. adam returns ranked code results to the LLM, which generates the conversational explanation shown here.
adam runs fully offline or within your private infrastructure. Natural-language queries return ranked code locations without any external calls. Optionally connect an LLM for conversational explanations—adam provides the structured context so the model reasons from real code, not guesses.
All indexing, analysis, and search runs on your machine or within your private infrastructure. No external services required. Zero data transmitted unless you choose to connect an LLM.
All processing happens locally. No API calls to external AI services required for core functionality.
Enterprise Plus includes full air-gapped deployment. Zero network access required.
Automatically find patterns, problems, and security issues across your entire codebase.
Automatically detects design patterns, near-duplicates, copy-paste suspects, structural outliers, refactoring candidates, and cross-language patterns across your codebase.
Detected patterns:
Repository Pattern (12 occurrences)
src/repositories/*.ts
Factory Pattern (5 occurrences)
src/factories/*.ts
Singleton Pattern (3 occurrences)
src/services/database.ts
src/services/cache.ts
src/services/logger.ts
Comprehensive security scanning detects hardcoded secrets, injection vulnerabilities, XSS, path traversal, insecure deserialization, and more—with deep code understanding for precision.
Security scan results:
ERROR Hardcoded API key
src/config/api.ts:23
Use environment variables instead
WARNING SQL query interpolation
src/db/queries.ts:45
Use parameterized queries
NOTE No rate limiting
src/api/routes/auth.ts:12
Consider adding rate limits
Detects complexity hotspots, code smells, and maintainability issues. Plus style analysis for consistency across your codebase.
Index multiple branches simultaneously. Search across feature branches without switching context.
Authentication, access control, audit logging, and enterprise-grade encryption—built for security-first organizations.
Single sign-on with SAML and OAuth2/OIDC support for enterprise identity providers.
Role-based access control with admin and user roles. Organization-scoped resource management and permissions.
Complete audit trail for device registration, access events, and admin actions. Integrates with your existing monitoring.
Enterprise-grade encryption at rest with per-device keys and customer-managed master keys. FIPS 140-2 compliant.
Per-user device limits with secure token management. Configurable expiry policies and instant server-side revocation.
Automated subscription management with seamless upgrades, downgrades, and a 14-day grace period after cancellation.
Fast indexing. Instant search. Minimal setup. Get productive in minutes.
Average search time
Languages supported
Typical setup time
Files per repo supported
See how adam can transform your development workflow.