Phases are the execution units of a Guided Generative Workflow (GG) project.Documentation Index
Fetch the complete documentation index at: https://docs.joggr.ai/llms.txt
Use this file to discover all available pages before exploring further.
/gg-plan creates them as the output of the project-level planning stage — by the time a phase exists, the discussion, research, and plan have already happened at the project level. Each phase then moves through its own cycle independently.
Phase Lifecycle
Scaffold
The phase file is created by
/gg-plan with its task list, file targets, and acceptance
criteria. The phase exists but no work has started.Execute
/gg-execute groups the phase’s tasks into waves and runs them in parallel. Progress is
checkpointed after each wave.Review
Use
/gg-review to open the phase plan in a local review UI
where you can leave inline annotations and approve or request changes before execution begins.If a phase needs more context before execution, you can re-run
/gg-discuss, /gg-research, or
/gg-plan against it by passing the phase number — for example, /gg-plan 2 to refine phase 2’s
task breakdown.Tracking Phase State
Each phase tracks two values that together describe its current state:| Concept | Values | Description |
|---|---|---|
| Step | scaffold, discuss, research, plan, execute, verify | Which stage of the lifecycle the phase is in. Moves forward as you advance through the pipeline. |
| Status | pending, running, done, error, skip | Whether the current step has completed. Resets to pending or running each time the step advances. A phase with skip status is excluded from execution. |
plan/done— planning is complete, the phase is ready for executionexecute/running— tasks are currently being executedverify/done— the phase has passed verification and is completeexecute/error— execution failed and needs attention before continuing
step and status fields at the top of the file reflect the current state.
Dependencies
A phase can depend on other phases. When a phase has dependencies, it is blocked from execution until all of its dependencies reachverify/done or are explicitly skipped.
Dependencies are defined in plan.md during the planning stage. They represent ordering constraints — for example, a database schema phase must reach verify/done before an API phase that relies on that schema can begin execution.
Workflows enforces dependencies automatically. When you run /gg-execute, it checks whether the phase’s dependencies are satisfied before proceeding. If a dependency is not met, it tells you which phases need to complete first.
Task Structure
Each phase contains a list of tasks — the atomic units of work Workflows executes. Each task has an ID, description, file targets, a check command, and acceptance criteria. See the /gg-plan reference for the full field breakdown.Wave Grouping
During execution,/gg-execute groups tasks into waves based on their file targets:
- Tasks that touch different files run in the same wave (parallel)
- Tasks that touch the same files run in different waves (sequential)
- Task dependencies from the plan are also respected — if task X depends on task Y, X runs in a later wave
Task Status
Each task tracks its own status independently:| Status | Description |
|---|---|
todo | Not yet executed |
done | Completed successfully |
skip | Skipped by user choice |
phase.md file alongside each task entry.
Phase Files
Each phase is stored in its own directory under the project:/gg-plan and updated by /gg-execute and /gg-verify as the phase progresses.