Fleetworks · Deployments and agentic workloads
One control plane for what you ship and the agents that help
Applications, pipelines and deployments on one side; agents, multi-agent processes and MCP servers on the other. One organization model, one set of access bits on every resource, one audit trail, and a REST API over all of it.
Two sides, one model
Every governed resource belongs to one Organization and carries an Owner user, an Owner group and a Mode. The same check applies whether the resource is an Application or an Agent.
- 01
Applications
One deployable service, with Quotas per Cluster, Namespaces, Ownership links, Extended resources and an LDAP-group Permission list.
- 02
Pipelines and Pipeline runs
The CI/CD definition for an Application, and each execution of it from NOT_STARTED to SUCCEEDED, FAILED or CANCELED.
- 03
Deployments and Stages
A running instance of an Application on a Cluster, with service type and DNS, and the Stages it moves through.
- 04
Agents and Agent runs
An AI worker defined by kind, Runtime, model, instructions and the MCP servers it may use. Each Invoke records an Agent run with its token usage.
- 05
Processes and Approval gates
A multi-agent template of ordered Steps under a sequential, concurrent, handoff or supervisor Topology. An Approval gate pauses the Process run until a person decides.
- 06
MCP servers
A governed registry of Model Context Protocol tool servers that Agents may use. Secret headers are stored but never shown back.
What Helmsman does
- 01
Application lifecycle
Create, configure and delete Applications with Quotas per Cluster, Namespaces, Ownership links and Extended resources.
- 02
Pipelines and Pipeline runs
Record the CI/CD definition of each Application and every execution of it, with Commit statuses beside them.
- 03
Deployment tracking
Follow each Deployment of an Application on a Cluster, with service type, DNS, Stages and Events.
- 04
Clusters and Quotas
List and inspect the Kubernetes Clusters Deployments run on, and hold each Application’s CPU, memory and jobs allowance per Cluster.
- 05
Agents as governed resources
Define AI workers with a kind, Runtime, model, instructions and allowed MCP servers, then Invoke them and audit each Agent run.
- 06
Pluggable Runtimes
Agents run through a framework-agnostic Runtime port, picked per Agent, with no silent fallback to a billed Runtime.
Public records are one request away
Most routes need a Zitadel-issued bearer token or a Service account PAT. Cluster lists, Commit statuses and Search are readable without one, and Search returns only resources whose Mode lets everyone read them. In the examples, HELMSMAN_API is the API host you were given.
# public records, no credentials
curl "$HELMSMAN_API/api/v1/search?q=deploy"
# with a Service account PAT
curl -H "Authorization: Bearer wl_…" $HELMSMAN_API/api/v1/applicationsFrom an empty organization to governed work
- Step 1
Sign in
Sign in through Zitadel, the shared Fleetworks identity provider. Your Org role decides what you may do inside your Organization.
- Step 2
Record applications
Create an Application, set its Quotas per Cluster and its Namespaces, and link its Pipelines.
- Step 3
Track delivery
Pipeline runs, Deployments, Stages, Events and Commit statuses arrive through the API, from people or from CI with a Service account PAT.
See every deployment and every Agent run in one place
Sign in with your Fleetworks account, or start with the public endpoints of the API.