nitro agent tasks

The tracker is the single source of truth for work and decisions, because sessions die and context compacts while tasks survive. The tracker never runs git commands and never writes anything to the working tree, so none of its state can land in a commit. Always call a subcommand, since bare nitro agent opens an interactive TUI and blocks the session. If nitro itself is not found, the CLI is not installed — stop and tell the user to install it: https://chillicream.com/docs/nitro/cli/installation. Do not attempt to install it yourself.

Core principles

• Use --output json and take ids from results. Ids are never constructed. List commands return compact snapshots; nitro agent tasks show <id> --output json returns full state -- description, comments, dependencies, blockers.
• Comments are the decision log. User rulings and binding decisions go in comment add, not only in descriptions; agents read them through show.
• Pass your actor name on every write: --actor <name> on create, q, update, close, reopen, delete, defer, undefer, dep add, dep remove, comment add, label add, label remove, and epic close-eligible. Read commands take no actor. The name is handed to you, never invented. With Nitro's hooks installed, the session-start hook states it in your context: Your Nitro actor name is "maya". Otherwise nitro agent login allocates one and prints it.
• Close with evidence. close --reason names the commits or facts that prove it; when unsure, comment instead of closing.

Work a task