Skip to main content

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-task runs the full discuss-research-plan-execute-verify pipeline in one session. No project scaffolding, no phase files — just a description, a flat task list, and working code at the end.

​
Prerequisites

Complete Setup Workflows first — it covers CLI install, authentication, repo initialization, and codebase doc generation.

​
1. Start the task

In Claude Code or Cursor, run: 
/gg-task
Describe what you want to do when prompted. Be specific — the more context upfront, the better the research and plan. 
Fix the off-by-one error in the pagination component — the last page
shows an empty state instead of the final results.
You can also pass a description directly to skip the prompt: 
/gg-task fix the off-by-one error in pagination
Workflows echoes the description back, asks you to confirm a slug (e.g., fix-pagination-off-by-one), and scaffolds spec.md under .joggr/.gg/tasks/fix-pagination-off-by-one/.

​
2. Complete the discussion

Workflows runs a short structured interview covering scope, constraints, edge cases, and acceptance criteria. It asks 2-3 questions at a time and moves on once each area is covered. Say “move on” or “done” at any point to end the discussion early. When finished, it flows directly into research — no separate command needed.

​
3. Review research, then approve the plan

After research completes, Workflows checkpoints spec.md and opens a review UI. Read the research findings and acceptance criteria — if anything looks off, edit spec.md directly before continuing.
  • Approve — planning starts immediately
  • Save and stop — exits; resume later with /gg-task fix-pagination-off-by-one
  • Skip review — proceeds without opening the UI
Once planning finishes, the same review gate opens for plan.md. This is your main chance to catch scope issues — wrong files, tasks that are too broad, missing steps — before execution begins. If the plan needs changes, say so before approving and Workflows will revise it.

​
4. Monitor execution

Workflows groups non-conflicting tasks into waves and runs them in parallel. After each wave it checkpoints the changes and asks whether to continue or pause. Run /gg-status at any point to see which tasks are done and what’s still in progress.
/gg-task saves your work automatically — after research, after planning, after each wave, and after verification. You can safely close the session at any wave boundary and resume later with /gg-task your-task-slug. If the session is getting long and output quality is dropping, choose Clear + Continue at the next wave prompt — it checkpoints the wave, then you run /clear and /gg-task your-task-slug to resume with a fresh context.

​
5. Archive when done

After verification passes, Workflows offers to archive the task: 
jog gg tm archive fix-pagination-off-by-one
Archiving zips the task directory to .joggr/.gg/.archive/, deletes the source directory, and clears it from /gg-status.

​
Troubleshooting

If multiple tasks exist, Workflows may not know which one you mean. Pass the slug explicitly: /gg-task your-task-slug. Run jog gg tm list to see all active task slugs.
Workflows needs codebase documentation to research and plan effectively. Run /gg-codebase to generate it. This only needs to be done once per repository — re-run if your codebase structure changes significantly.
Re-invoke with the same slug — /gg-task your-task-slug — and Workflows picks up where you left off. Completed steps are skipped entirely. Steps interrupted mid-run are retried from the beginning, except execution — which resumes from the last incomplete task.
Workflows offers three options: fix and re-verify, accept as-is with an admonition in spec.md, or go back to execution to address specific tasks.