For Agents
Register services and endpoints in a central registry, then resolve them by name for clients running across GCP and hybrid environments.
Use for: Register a new microservice with its endpoints in Service Directory, Resolve all healthy endpoints for the 'orders' service, List every service registered in the production namespace, Add a new endpoint with a specific port to an existing service
Not supported: Does not handle health checking, traffic routing, or DNS resolution itself — use for service registration, endpoint metadata, and resolve lookups only.
Service Directory is Google Cloud's managed service registry for discovering, publishing, and connecting services across GCP, on-prem, and other clouds. The API organises services into namespaces, registers individual services with metadata, and tracks endpoints with addresses, ports, and annotations. A resolve operation returns all healthy endpoints for a service so clients can do client-side load balancing or DNS lookup.
Install Jentic One Beta
Jentic One is a self-hosted execution layer for AI agents. It lets your agent call the Service Directory API, or any other public or private API you need. You set the rules, the agent never sees your credentials, and every call is logged.
Two steps, two machines. Install the instance in a safe environment, then register your agent from wherever it runs.
Step 1: Jentic One Host machine
# On the machine that will host your Jentic One instance:
curl -fsSL https://raw.githubusercontent.com/jentic/jentic-one/main/tools/install.sh | shStep 2: Agent machine
# On the machine where your agent runs (keep this separate from the instance):
curl -fsSL https://raw.githubusercontent.com/jentic/jentic-one/main/tools/install.sh | sh
jentic register # connects your agent to your Jentic One instanceJentic One is in public beta. The setup above keeps your agent separate from the instance, which is what you want before using real credentials: an agent running as the same OS user as Jentic One can read its stored keys directly. Just evaluating? A single local install is fine to start. See the secure deployment guide for the tiers.
What an agent can do with Service Directory API.
Create, list, and delete namespaces that group services within a region
Register, update, and delete services with arbitrary metadata annotations
Add, list, and remove endpoints (address, port, annotations) for each service
Resolve a service to retrieve all of its current endpoints in one call
Filter and search services or endpoints with a flexible filter expression
Manage IAM bindings on namespaces, services, and endpoints
Patterns agents use Service Directory API for, with concrete tasks.
★ Hybrid Service Registry
Use Service Directory as a single registry across GCP, on-prem, and other clouds. Services running in any environment register themselves with their network address and port, and clients resolve by service name regardless of where the workload lives. Pairs with Cloud DNS to expose entries through DNS and with VPC private connectivity for hybrid endpoints.
Call POST /v1/{+parent}/services/{serviceId}/endpoints with address, port, and annotations to register a new instance, then call services:resolve to confirm it appears.
Microservice Discovery for Internal Apps
Internal apps and serverless workloads call services:resolve at startup to discover the addresses of dependent services. Service Directory tracks endpoint metadata so clients can pick by region, version, or any custom annotation, and updates propagate quickly when endpoints register or de-register.
Call POST /v1/{+name}:resolve on the service to retrieve every endpoint, then pick one matching annotation 'version=v2'.
Decommissioning Audit
Run periodic audits that list every service in every namespace, identify ones with no recent registrations or no healthy endpoints, and flag them for cleanup. The API supports filter expressions on labels, regions, and last-update timestamps to make this kind of housekeeping straightforward.
List namespaces, then for each namespace call services.list with filter='annotations.tier="critical"' and assert at least one endpoint is registered.
AI Agent Topology Mapper
An AI agent that reasons about a deployment's topology pulls services and endpoints from Service Directory to build a live dependency map. Through Jentic the agent searches by intent and never holds the GCP service account credentials directly, so the same agent can run across multiple GCP organizations safely.
List every namespace, list every service in each namespace, list every endpoint per service, and emit a graph of service-to-endpoint relationships.
14 endpoints — service directory is google cloud's managed service registry for discovering, publishing, and connecting services across gcp, on-prem, and other clouds.
METHOD
PATH
DESCRIPTION
/v1/{+parent}/namespaces
List namespaces in a project and region
/v1/{+parent}/services
Register a new service
/v1/{+parent}/endpoints
List endpoints for a service
/v1/{+parent}/endpoints
Register a new endpoint
/v1/{+name}:resolve
Resolve a service to its endpoints
/v1/{+name}
Delete a namespace, service, or endpoint
/v1/{+parent}/namespaces
List namespaces in a project and region
/v1/{+parent}/services
Register a new service
/v1/{+parent}/endpoints
List endpoints for a service
/v1/{+parent}/endpoints
Register a new endpoint
/v1/{+name}:resolve
Resolve a service to its endpoints
/v1/{+name}
Delete a namespace, service, or endpoint
Three things that make agents converge on Jentic-routed access.
Credential isolation
GCP service account credentials and OAuth tokens are stored encrypted in the Jentic MAXsystem vault. Agents receive a short-lived scoped access token at execution and never see the underlying service account keys, even when the same agent works across multiple GCP organizations.
Intent-based discovery
Agents search Jentic by intent (e.g. 'resolve a service to its endpoints') and Jentic returns the matching services:resolve operation with its full input schema, so the agent does not have to navigate Service Directory's resource hierarchy by hand.
Time to first call
Direct integration: 1-2 days to set up namespaces, IAM bindings, and a resolver client with caching. Through Jentic: under 30 minutes — search, load, execute.
Alternatives and complements available in the Jentic catalogue.
Specific to using Service Directory API through Jentic.
What authentication does the Service Directory API use?
Service Directory uses OAuth 2.0 with the cloud-platform scope and IAM permissions like roles/servicedirectory.editor on the namespace, service, or endpoint. Through Jentic the credentials are stored encrypted in MAXsystem and a scoped token is provided at execution.
Can I use Service Directory for on-prem services?
Yes. Endpoints record arbitrary address and port values, so you can register on-prem or third-cloud services and resolve them from clients running in GCP or via Cloud DNS forwarding zones.
What are the rate limits for the Service Directory API?
Google enforces a default per-project quota of 600 read requests per minute and 60 write requests per minute on the registry resources. Resolve calls have higher per-region limits because they sit on the discovery hot path.
How do I resolve a service through Jentic?
Run the Jentic search query 'resolve service directory endpoints', load POST /v1/{+name}:resolve, and execute it with the service resource name. Jentic injects the GCP credentials so the agent only needs the service name and an optional max-endpoints filter.
Does Service Directory health-check endpoints?
No. Service Directory stores the endpoints you register; health checking is the responsibility of the registering workload or a fronting load balancer. Clients should still validate health on their side after resolving.
Can I expose Service Directory entries through DNS?
Yes. Cloud DNS supports a Service Directory zone type that mirrors a namespace as a private DNS zone. Updates to endpoints propagate to the DNS view automatically without separate record management.
GET STARTED