Honeydipper is an event-driven, policy-based orchestration system that is tailored towards SREs and DevOps workflows, and has a pluggable open architecture. The purpose is to fill the gap between the various components used in DevOps operations, to act as an orchestration hub, and to replace the ad-hoc integrations between the components so that all the integrations can also be composed as code.
DipperCL stands for Dipper Control Language. It is a yaml based language used for configuring Honeydipper, and defining assets within Honeydipper.
Honeydipper uses 4 type of assets. They are driver, system, workflow and rules.
Honeydipper driver is a golang program that is dynamically loaded by the daemon to perform various tasks or ingest events. For example, webhook driver, web driver, kubernetes driver, etc.
A system is an abstract representation of a physical system in Honeydipper. It is defined by a group triggers and functions through which the system interacts with Honeydipper and the world. For example, github system has a group of triggers implemented through webhook driver, and a group of functions that calls github APIs through web the driver.
A workflow is a data structure that defines the work needed to complete a task. It can contains steps, threads or invoking functions, other workflows. It can also contain definition of context variables, and evaluate conditions before taking actions.
A rule defines a trigger, a set of conditions and a workflow indicating that the workflow needs to be executed when the trigger is fired and the conditions are met.
Honeydipper supports three types of drivers:
See the Remote Driver And Registry Guide for detailed instructions on configuring remote drivers, registries, and source policies.
Yes, you can use a local binary as a driver by configuring it as a remote driver with localPath in the handlerData. You will also need to enable the local source policy.
Honeydipper v4 introduces several new features:
An Agent is an AI-powered component that can execute workflows, call tools, and engage in conversations. Agents can be configured with different AI models (OpenAI, Gemini, Ollama), system prompts, and tool sets. They support multi-turn conversations, streaming responses, and sub-agent delegation.
Agents are configured under the agents section in your configuration. Each agent has a name, driver (AI model), system prompt, and optional tools. See the Configuration Guide for detailed examples.
The Honeydipper API is an HTTP REST API that allows you to interact with the daemon programmatically. It supports endpoints for managing events, agent conversations, secrets, and user profiles. The API uses JWT authentication and supports SAML and GitHub OAuth for user authentication.
The API supports multiple authentication methods:
Authorization: Bearer <JWT> header with a valid JWT token./auth/saml/* endpoints for SAML-based authentication./auth/github/callback endpoint for GitHub OAuth.Honeydipper can be started by running the daemon binary with the appropriate configuration. The daemon will load the configuration, initialize the services, and start processing events.
honeydipper -c /path/to/config.yaml
Honeydipper automatically reloads the configuration every 60 seconds by checking the configured git repositories for changes. You can also trigger a reload by sending a broadcast:reload message through the event bus.
Use the configcheck mode to validate your configuration without starting the daemon:
honeydipper -configcheck -c /path/to/config.yaml
You can trigger a workflow manually by sending an event through the API or by using the job mode to run a workflow once and exit:
honeydipper -job -c /path/to/config.yaml
Honeydipper supports encrypted secrets using the ENC[...] prefix and secret lookups using the LOOKUP[...] prefix. See the Interpolation Guide for detailed examples.
Common reasons for driver startup failures:
builtin, remote, null) is correctly configured.registry, direct, local).requiredPackages, ensure the required packages are defined for your package manager.You can debug workflow execution by:
export feature to inspect workflow context data.Honeydipper has a rollback mechanism that restores the last known good configuration if a reload fails. Check the logs for errors and validate your configuration using configcheck mode.
Please open an issue on the Honeydipper GitHub repository with a detailed description of the bug or feature request.