| name | kurrent-docs |
| description | Use when the user is working with KurrentDB, from writing application code against the client SDKs to operating the self-hosted server or running on Kurrent Cloud. |
Kurrent Documentation
Match the user's need to the row, then take the action. Load only the files the action names. Don't pull in a whole section. The actual answers live in the linked file, not in this index.
Route top to bottom: first check whether a sibling skill owns the request, then pick a domain, then load the file. Don't backtrack.
Does a sibling skill own this?
Some requests belong to a more specialized skill. Check these first; defer if one matches, otherwise route within this skill below.
| User need | Defer to |
|---|
"My app can't connect" triage, or tuning the client (connection-string parameters, nodePreference, keepalive, deadlines, serverless reuse) | kurrentdb-connection. This index has the raw parameter reference; that skill has the opinionated guidance on what to set. If the failure is server-side (cert, ACL, gossip), also load the Database or Cloud files named below. |
| Querying events with SQL (ad-hoc in the UI or via Arrow Flight SQL), or creating a user-defined index to query/filter by a payload field | kurrentdb-index-queries. It owns the kdb.records / usr."<index>" SQL surface, Flight SQL client setup, and user-defined index creation. |
| Migrating or upgrading a client SDK or server (legacy TCP to gRPC, EventStoreDB to KurrentDB rebrand, version upgrades) | kurrent-upgrade. |
Pick a domain
| User need | Action |
|---|
Application code calling KurrentDB from .NET, Java, Node.js, Python, Go, or Rust: AppendToStream, SubscribeToAll, EventData, catch-up vs persistent subscriptions, gRPC errors | Go to Client SDKs |
The self-hosted KurrentDB server itself: install/upgrade, config flags, running on Kubernetes via the Operator, TLS, ACLs, /admin/logs, /metrics, scavenge, projections, indexes, HTTP API, connectors (Kafka/Mongo/SQL/…) | Go to Database (self-hosted) |
Kurrent Cloud: console.kurrent.cloud, VPC/VNet peering, Tailscale, K8s connectivity, sizing, backups, Cloud-side Terraform/Pulumi, CloudWatch/Opsgenie/Slack integrations | Go to Cloud |
Client SDKs
Files live at references/client-sdks/<lang>/<topic>.md. Pick one language (dotnet, java, nodejs, python, go, rust) and load only the topics the question actually touches. Don't load multiple languages for the same topic. For any code answer, follow the rules under Writing answers.
| User need | Action | Available in |
|---|
First connection, package install, building a KurrentDBClient | Read <lang>/getting-started.md | all 6 |
AppendToStream, expected revision, optimistic concurrency, idempotent writes | Read <lang>/appending-events.md | all 6 |
Single stream or $all, forward/backward, reading from a revision | Read <lang>/reading-events.md | all 6 |
| Catch-up subscriptions, live tail, checkpointing the last-seen position | Read <lang>/subscriptions.md | all 6 |
| Server-managed consumer groups, ack/nack, parking | Read <lang>/persistent-subscriptions.md | all 6 |
| Soft delete vs hard delete (tombstone), truncate-before | Read <lang>/delete-stream.md | all 6 |
| Managing/reading projections from the client SDK | Read <lang>/projections.md | dotnet, java, nodejs, python, go (not rust) |
| Username/password, certificates, per-call user credentials | Read <lang>/authentication.md | dotnet, java, nodejs, go, rust |
| OpenTelemetry instrumentation, traces, metrics | Read <lang>/observability.md | dotnet, java, nodejs, python |
Parsing/building kurrentdb:// and esdb:// URIs | Read <lang>/connection-strings.md | python (other langs cover this inline in getting-started.md) |
Database (self-hosted)
Files live at references/database/. This is everything about the KurrentDB server process: binary, configuration, security, diagnostics, operations, HTTP API, and the server-managed features (projections, indexes, queries, connectors).
Install, upgrade, release info
| User need | Action |
|---|
| Installing the server binary (Docker, package managers, tarballs) | Read quick-start/installation.md |
| Data, log, and config paths by platform | Read quick-start/default-directories.md |
| Upgrading between versions, breaking-change pre-flight | Read quick-start/upgrade-guide.md |
| Recent feature additions | Read quick-start/whatsnew.md |
| Per-version changelog | Read release-schedule/release-notes.md |
| Support windows, EOL dates | Read release-schedule/previous-versions.md |
| What the binary phones home, opting out | Read usage-telemetry.md |
Kubernetes Operator
Running the self-hosted server on Kubernetes via the KurrentDB Operator (an Enterprise-only feature). Files live under kubernetes-operator/, tracking the latest Operator release.
| User need | Action |
|---|
| Installing the Operator via Helm, OpenShift/OperatorHub, or OLM | Read kubernetes-operator/getting-started/installation.md |
CRD/spec reference for KurrentDB, KurrentDBBackup, KurrentDBBackupSchedule | Read kubernetes-operator/getting-started/resource-types.md |
| Deploying a cluster: production settings, secure/insecure examples, read-only replicas, archiver | Read kubernetes-operator/operations/database-deployment.md |
| Changing a live cluster: image/CPU/memory/disk/replica updates, scaling, reload/restart triggers | Read kubernetes-operator/operations/modify-deployments.md |
Backups and restore: volume snapshots, scheduled backups, retention, restoring via sourceBackup | Read kubernetes-operator/operations/database-backup.md |
| TLS certificates: cert-manager, LetsEncrypt, FQDN templates, wiring certs via secrets | Read kubernetes-operator/operations/managing-certificates.md |
Advertised addresses and traffic strategy (ServiceName/FQDN/SplitDNS), network security | Read kubernetes-operator/operations/advanced-networking.md |
Configuration
| User need | Action |
|---|
| How config works: file vs env vars vs CLI, precedence rules | Read configuration/configuration.md |
| Storage layout, chunks, indexes/cache, scavenge tuning | Read configuration/db-config.md |
| Gossip seeds, leader election, cluster sizing | Read configuration/cluster.md |
| Ports, interface binding, advertise hosts | Read configuration/networking.md |
Security
| User need | Action |
|---|
| TLS for client/internal traffic, FIPS, certificate chains | Read security/protocol-security.md |
| Auth provider config, anonymous access toggles | Read security/security-options.md |
| Built-in user store, internal vs external auth | Read security/user-authentication.md |
ACLs, $admins/$ops groups, default ACL | Read security/user-authorization.md |
Diagnostics & monitoring
| User need | Action |
|---|
/admin/logs, log structure, levels, sinks | Read diagnostics/logs.md |
/metrics Prometheus endpoint, exact metric names | Read diagnostics/metrics.md |
| Exporters/dashboards (Grafana, Datadog, …) | Read diagnostics/integrations.md |
| What to alert on, SLI selection | Read diagnostics/best-practices.md |
Operations
| User need | Action |
|---|
| Filesystem-level backups and restore procedure | Read operations/backup.md |
| Manual scavenge, when it's needed and what it does | Read operations/scavenge.md |
| Scheduling scavenge across the cluster | Read operations/auto-scavenge.md |
| Rotating TLS certs without downtime | Read operations/cert-update.md |
| Erasing event payloads (GDPR / right-to-be-forgotten) | Read operations/redaction.md |
HTTP API
| User need | Action |
|---|
| When to use HTTP vs gRPC, supported operations | Read http-api/introduction.md |
| Endpoints, request/response formats | Read http-api/api.md |
ES- / Kurrent- headers: expected version, event IDs | Read http-api/optional-http-headers.md |
| Driving persistent subscriptions over HTTP | Read http-api/persistent.md |
| HTTP auth: Basic, JWT, certs | Read http-api/security.md |
Server features
| User need | Action |
|---|
| Embedded Web UI, what each page does | Read features/admin-ui.md |
Stream semantics: $all, system streams, max age/count, metadata | Read features/streams.md |
| Offloading old chunks to cold storage | Read features/archiving.md |
| Server-side view of persistent subs (queue model, settings) | Read features/persistent-subscriptions.md |
Projections (server-side)
| User need | Action |
|---|
| What projections are, system vs user | Read features/projections/intro.md |
$by_category, $by_event_type, $streams, etc. | Read features/projections/system.md |
| Writing JS projections, partitioning | Read features/projections/custom.md |
| Enabling, threading, checkpoint behaviour | Read features/projections/settings.md |
| The new projection engine and migration path | Read features/projections/engine-v2.md |
| End-to-end walkthrough | Read features/projections/tutorial.md |
Indexes & queries
| User need | Action |
|---|
| The built-in PTable/MemTable index | Read features/indexes/default.md |
| Secondary indexes, when and why | Read features/indexes/secondary.md |
Defining your own indexes (creating/querying one routes to kurrentdb-index-queries) | Read features/indexes/user-defined.md |
Query browser inside the admin UI (SQL workflow routes to kurrentdb-index-queries) | Read features/queries/ui.md |
Apache Arrow Flight SQL endpoint (client setup routes to kurrentdb-index-queries) | Read features/queries/flightsql.md |
Connectors
KurrentDB's built-in source/sink connectors, managed by the server itself, configured per stream. Files at references/database/features/connectors/. Start with intro.md for the model, then load the one sink/source file for the system the user is wiring up.
| User need | Action |
|---|
| What connectors are, source vs sink model | Read intro.md |
| Filtering, transformations, retries, DLQ | Read features.md |
| Creating/starting/stopping connectors via API or UI | Read manage.md |
| Common config keys: filter, checkpointing, parallelism | Read settings.md |
| Per-connector metrics and health signals | Read metrics.md |
| Forward events to Kafka | Read sinks/kafka.md |
| Write events to MongoDB | Read sinks/mongo.md |
| Write events to a SQL database | Read sinks/sql.md |
| Write events to Elasticsearch | Read sinks/elasticsearch.md |
| Generic HTTP / webhook sink | Read sinks/http.md |
| Forward events to Apache Pulsar | Read sinks/pulsar.md |
| Forward events to RabbitMQ | Read sinks/rabbitmq.md |
| Stream events into a Serilog-compatible logger | Read sinks/serilog.md |
| Ingest from a Kafka topic into a KurrentDB stream | Read sources/kafka.md |
| Ingest from inbound HTTP webhooks | Read sources/webhook.md |
Cloud
Kurrent Cloud. Files at references/cloud/.
Orientation
| User need | Action |
|---|
| What Cloud is, regions, organization/project model | Read introduction.md |
| Common pre-sales / pre-onboarding questions | Read faq.md |
Getting started & networking
| User need | Action |
|---|
| Standing up a cluster on a public network | Read getting-started/public.md |
| AWS PrivateLink / VPC peering to a Cloud cluster | Read getting-started/private-access/aws.md |
| Azure Private Link / VNet peering | Read getting-started/private-access/azure.md |
| GCP Private Service Connect / VPC peering | Read getting-started/private-access/gcp.md |
| Public-network connectivity model | Read networking/public-network.md |
| Private-network connectivity model | Read networking/private-network.md |
| Connecting a Kubernetes workload to a Cloud cluster | Read guides/kubernetes.md |
| Tailscale-based access | Read guides/tailscale.md |
Operations
| User need | Action |
|---|
| Picking instance class, capacity planning | Read ops/sizing.md |
| Cloud-managed backups, restore, retention | Read ops/backups.md |
| The "Jobs" model for long-running cluster operations | Read ops/jobs.md |
| Cluster lifecycle event feed | Read ops/events.md |
| SSO, MFA, org/project roles | Read ops/account-security.md |
| Migrating from self-hosted to Cloud | Read guides/migration.md |
Infrastructure as Code
| User need | Action |
|---|
kurrent-io/kurrentcloud Terraform provider | Read automation/terraform.md |
| Pulumi provider | Read automation/pulumi.md |
Cloud-side integrations
| User need | Action |
|---|
| Forwarding metrics/logs to AWS CloudWatch | Read integrations/cloudwatch.md |
| Alert routing to Opsgenie | Read integrations/opsgenie.md |
| Slack notifications for cluster events | Read integrations/slack.md |
Same word, two domains
Some terms map to a client-SDK file or a server/cloud file depending on whether the user is writing app code or operating the cluster. Disambiguate by that, then route.
| User need | App code (Client SDKs) | Server / operations |
|---|
| "Subscribe to events" | <lang>/subscriptions.md (catch-up), <lang>/persistent-subscriptions.md (consumer groups) | As a server-side bridge into Kafka/Mongo/SQL/etc., Database features/connectors/ |
| "Projections" | <lang>/projections.md (read/create from the client) | Configuring server projections, JS bodies, engine tuning: Database features/projections/ |
| "Persistent subscriptions" | <lang>/persistent-subscriptions.md (consume from one) | Operating the server side (groups, parking, settings): Database features/persistent-subscriptions.md |
| "Indexes" | (none; see kurrentdb-index-queries for user-defined) | What they are server-side (storage, build, config): Database features/indexes/secondary.md |
| "Provision a cluster" | (none) | Managed: Cloud getting-started/ + automation/. Self-hosted: Database quick-start/installation.md + configuration/cluster.md. On Kubernetes: Database kubernetes-operator/ |
A request can genuinely span domains (e.g. a .NET app reading from a peered Cloud cluster). Load files from both in parallel; the tables don't overlap, so it's cheap.
Writing answers
Code answers must be runnable. Topic files show only the operation and assume the setup above them, so for any code request also load <lang>/getting-started.md and assemble a self-contained snippet: imports, client construction from a connection string, and an inline definition of any payload type you reference (with the getters/fields serialization needs). Never return a fragment with an undefined client or payload class.
Copy exact names from the reference file, don't reconstruct them from memory. Kurrent rebranded from EventStore and the names mutated in non-obvious ways:
EventStore.Client → KurrentDB.Client (the .NET package; equivalents in every language)EventStoreDB → KurrentDB (server binary, docs, client class names)EventStore/eventstorecloud → kurrent-io/kurrentcloud (Terraform provider)esdb:// → kurrentdb:// (connection-string scheme on v25+; esdb:// still works when either side predates the rename)
A package-manager coordinate is not the import path. Take the namespace, package, module, or use path verbatim from the reference file's own example code; never reconstruct it from the install coordinate. The two routinely differ across every language (e.g. Java installs io.kurrent:kurrentdb-client but imports io.kurrent.dbclient.*; .NET's KurrentDB.Client package and namespace happen to match, others don't). If the reference shows the install line but no import, the import still comes from the example, not the artifact id.
