Agent Name Service
github.com/ruvnet/agent-name-service- Category
- Cybersecurity
- Rank
- No. 999Tools index
- Pricing
- Open Source
- Type
- TOOL
- Builder
- ruvnet
- GitHub
- 81 stars
- Date
About
ANS Protocol from the OWASP GenAI Security Project — secure discovery and interaction framework for AI agents across organizations.
What it does
Agent Name Service is a TypeScript library for registering named agents, storing their metadata, and looking them up later. Registration creates an identity certificate, runs a security assessment, formats an agent card, and saves it in SQLite. It can also turn supplied tool metadata into an MCP manifest.
Why it's ranked here
This is a useful architectural prototype, but its security claims outrun the implementation. The registry combines validation, rate limiting, event logging, optional encrypted storage, certificates, and protocol formatting behind one service. However, the active threat analyzer always returns the same mock finding, and certificate verification checks structure rather than a real signature chain.
What's good
The code separates registry storage, certificate handling, protocol formatting, and threat reporting. Agent names receive length, character, and reserved-prefix checks. SQLite queries use parameters, stored cards can use authenticated encryption, and registrations can be limited by requester address. The package exposes both the main service and lower-level utilities for selective adoption.
Tradeoffs
Certificates only resemble X.509 records: private-key storage is a placeholder, signature validation accepts any well-formed certificate wrapper, and no trust chain is verified. The active security analysis is a fixed mock response rather than metadata-sensitive detection. Protocol formatters prepend labels to strings instead of producing demonstrated standards-compliant documents. Rate-limit state also lives only in process memory.
How to use it well
Use it as reference code or a prototype registry for a controlled TypeScript agent experiment. Its strongest workflow is registering metadata, inspecting the resulting threat report, persisting cards locally, and generating a basic MCP representation. Replace the certificate and analyzer components before production. It does not provide a deployed network API, mature public-key infrastructure, or proven cross-organization trust.
Technical notes+
mod.ts exports AgentNamingService, AgentRegistry, issueCertificate, formatAgentCard, and formatMCPManifest as a library surface. src/ans.ts constructs AgentRegistry with defaults, so registrations use in-memory SQLite unless the registry is instantiated separately with persistence options. It imports analysis from src/mastra-simple.ts, whose analyzeAgentSecurity returns a fixed mock report, rather than using src/mastra.ts. src/certificate.ts generates RSA keys but creates a simplified text certificate and treats a correctly wrapped certificate as signature-valid. src/db.ts supports AES-256-GCM encryption, parameterized queries, agent-name validation, and security-event storage. src/protocols.ts returns labeled strings rather than parsed A2A or MCP structures. package.json provides Jest scripts and declares TypeScript, sqlite3, uuid, and node-fetch dependencies.
Observed
- License
- MIT
- Primary language
- TypeScript
- Install surface
- Clone the repository and install dependencies with npm
- Interface
- Importable library service and utility exports; no CLI or network API is shown
- Storage
- SQLite, with in-memory storage by default and optional file persistence
- Package dependencies
- Runtime dependencies include sqlite3, uuid, and node-fetch
- Testing surface
- Jest scripts are provided for standard, watch, and coverage runs
Read from README.md, package.json, mod.ts, src/db.ts, src/ans.ts, src/mastra.ts, src/load-env.ts, src/protocols.ts, src/certificate.ts, src/test-mastra.ts, src/mastra-simple.ts, src/test-mastra-simple.ts, src/types/index.ts, src/core/ansService.ts, src/database/databaseService.ts.
What it can do
Discover AI agents across organizations
Organization or domain identifier → List of available AI agents with their capabilities
Authenticate AI agent identities
Agent credentials and security tokens → Verified agent identity and trust status
Route requests to appropriate AI agents
User request and target agent specification → Established connection to correct AI agent
Validate agent security compliance
Agent security configurations and policies → Security compliance report and recommendations
Register new AI agents in the network
Agent metadata, capabilities, and security credentials → Unique agent identifier and network registration confirmation
Monitor inter-agent communications
Agent interaction logs and communication streams → Security audit trail and interaction analytics
Tags
Tech Stack
Comments (0)
No comments yet
Editorially curated, with community endorsements as a secondary signal. Corrections welcome.