For Goose AI to be able to access Jira tickets you need an MCP Server. In this workflows we are using MCP server for Atlassian tools.
Secrets such as tokens are managed in the form of environment files. The templates for those files can be found in the ./templates directory. To configure the deployment, run make secrets first to copy the files to the .secrets directory where you can manually edit the files to add your tokens and more.
GOOGLE_API_KEY: take it from Google Cloud -> API & Services -> Credentials -> API Keys -> show key)
JIRA_PERSONAL_TOKEN: create PATs in your Jira/Confluence profile settings - usually under "Personal Access Tokens"
GITLAB_TOKEN with read permissions (read_user, read_repository, read_api). Note that some recipes require write access to Gitlab: use it at your own risk.
If you need to change the llm provider and model, they are stored in the Goose config file: goose-container/goose-config.yaml (GOOSE_PROVIDER, GOOSE_MODEL)
make build
Run the Jira MCP server from Atlassian and Goose separately, otherwise not all the input from your terminal is always redirected to the Goose container.
make run-mcp-atlassianmake run-goose(Requires enabling the Generative Language API or another LLM provider and configuring the environment variables as described in the configuration docs.)- Type List all In Progress issues at https://issues.redhat.com/projects/LD and wait for the output.
make clean
You can further manually run test and run the Goose recipes which are mounted into the container at /home/goose/recipes.
The recipes are defined in goose-recipes/. If you want to run goose-recipes/<recipe>.yaml, run the following:
make <recipe>make clean
This project uses pre-commit hooks. To set up:
pip install pre-commit
pre-commit install