harn-teamlisted

<!-- harnery:generated harn-team v=89c5b3c9; machine-owned; regenerated by `harn init`, removed by `harn deinit`. Edit the harnery template, not this file. --> Harnery gives you three ways to put more than one agent on a job. They differ by how long the objective outlives a single execution, and picking the wrong one is the usual reason a team spins. This skill owns that choice, the files a governed team needs, and the drive loop. ## Step 0: do not build a second team Run this before anything else, every time: ```bash harn governor list ``` If a goal already covers the objective, **you are done building** — drive the existing one (Step 3). A second team pointed at the same tree fights the first: both hold the same files and their frozen missions contradict each other. It is the most expensive mistake available here, and it looks like progress while you make it. Listings are tolerant: an unreadable record prints a warning row with its id, which is evidence the goal *exists* — inspect it before creating a replacement. Treat stale or suffixed experiment goals as shared runtime state; archiving them is an operator call, not yours. ## Step 1: pick the tier Match the tier to how long the objective must survive, not to how big it feels. | Surface | Use when | Dies when | |---|---|---| | `harn run <script>` | One bounded pass: deterministic JS stages fan out to headless subagents. | The pass ends. Nothing remembers it. | | `harn work create <title> <workflow>` | The objective