Plan with your agent
Open your repository in your usual coding agent. Use a checkout containing the setup files and skills BacklogGo installed. Pull those changes first if needed.
Talk through the feature with your agent. Agree on the scope and important assumptions before approving the plan. You can start with:
Use BacklogGo's installed planning skills to plan [my feature].
Ask me about important choices and assumptions before proposing a plan.
Keep the review summary short: what will change, what is out of scope, and decisions I need to make.
Wait for my approval before publishing tasks or marking any task agent-ready.Choose what may run
Ask your agent to apply these settings on the connected board:
| What you want | Status and label |
|---|---|
| Start a new task | Todo + agent-ready |
| Retry a task BacklogGo blocked | Keep Blocked, fix the reported problem, then re-add agent-ready. |
For a blocked retry, Todo also works after the problem is fixed and agent-ready is re-added. Moving a task alone does not authorize it.
These are the default status names; setup shows your project's names. Leave future work in Backlog or without agent-ready. Tasks wait for dependencies and available capacity.
Follow progress on your board. See Review and delivery for merge choices, or Troubleshooting if work stops.
If your planning agent needs help
It cannot find the installed skills
Ask it to read AGENTS.md and .agents/skills/backloggo-to-tickets/SKILL.md in your repository. You do not need to read the skills yourself.
It cannot create board tasks
Your separate coding-agent session needs provider access too. Ask it to check the repository's publication settings and board permissions. Keep credential values out of prompts and the repository.