Pre Workflow Hooks
Pre workflow hooks can be defined to run scripts before default or custom workflows are executed. Pre workflow hooks differ from custom workflows in several ways.
- Pre workflow hooks do not require the repository configuration to be present. This can be utilized to dynamically generate repo configs.
- Pre workflow hooks are run outside of Atlantis commands. Which means they do not surface their output back to the PR as a comment.
Pre workflow hooks can only be specified in the Server-Side Repo Config under the
pre-workflow-hooks do not prevent Atlantis from executing its workflows(
apply) even if a
run command exits with an error. This behavior can be changed by setting the fail-on-pre-workflow-hook-error flag in the Atlantis server configuration.
Atlantis Command Targetting
By default, the workflow hook will run when any command is processed by Atlantis. This can be modified by specifying the
commands key in the workflow hook containing a comma delimited list of Atlantis commands that the hook should be run for. Detail of the Atlantis commands can be found in Using Atlantis.
- id: /.*/
- run: ./plan-hook.sh
description: Plan Hook
- run: ./plan-apply-hook.sh
description: Plan & Apply Hook
commands: plan, apply
Dynamic Repo Config Generation
To generate the repo
atlantis.yaml before Atlantis can parse it, add a
run command to
pre_workflow_hooks. Your Repo config will be generated right before Atlantis parses it.
- id: /.*/
- run: ./repo-config-generator.sh
description: Generating configs
Customizing the Shell
By default, the command will be run using the 'sh' shell with an argument of '-c'. This can be customized using the
- id: /.*/
- run: |
echo "generating atlantis.yaml"
terragrunt-atlantis-config generate --output atlantis.yaml --autoplan --parallel
description: Generating atlantis.yaml
This is very similar to the custom workflow run command.
- run: custom-command
|Run a custom command
|Pre hook description
|The shell to use for running the command
|The shell arguments to use for running the command
runcommands are executed with the following environment variables:
BASE_REPO_NAME- Name of the repository that the pull request will be merged into, ex.
BASE_REPO_OWNER- Owner of the repository that the pull request will be merged into, ex.
HEAD_REPO_NAME- Name of the repository that is getting merged into the base repository, ex.
HEAD_REPO_OWNER- Owner of the repository that is getting merged into the base repository, ex.
HEAD_BRANCH_NAME- Name of the head branch of the pull request (the branch that is getting merged into the base)
HEAD_COMMIT- The sha256 that points to the head of the branch that is being pull requested into the base. If the pull request is from Bitbucket Cloud the string will only be 12 characters long because Bitbucket Cloud truncates its commit IDs.
BASE_BRANCH_NAME- Name of the base branch of the pull request (the branch that the pull request is getting merged into)
PULL_NUM- Pull request number or ID, ex.
PULL_URL- Pull request URL, ex.
PULL_AUTHOR- Username of the pull request author, ex.
DIR- The absolute path to the root of the cloned repository.
USER_NAME- Username of the VCS user running command, ex.
acme-user. During an autoplan, the user will be the Atlantis API user, ex.
COMMENT_ARGS- Any additional flags passed in the comment on the pull request. Flags are separated by commas and every character is escaped, ex.
atlantis plan -- arg1 arg2will result in
COMMAND_NAME- The name of the command that is being executed, i.e.
OUTPUT_STATUS_FILE- An output file to customize the success or failure status. ex.
echo 'failure' > $OUTPUT_STATUS_FILE.