Target selectors
List targets before using numeric selectors:
lictor -l
lictor -T -l
lictor -W -l
lictor -A -l
Indices start at one. Names and identifiers are case-sensitive. Herdr IDs are opaque; use the IDs returned by listing instead of predicting their counters.
| Mode | Selector | Meaning |
|---|---|---|
| Pane | N | Pane N in the baseline tab. |
| Pane | T.P | Pane P in tab T of the baseline workspace. |
| Pane | W:T.P | Pane P in tab T of workspace W. |
| Tab | N | Tab N of the baseline workspace. |
| Tab | W:N | Tab N of workspace W. |
| Workspace | N or a name | Workspace number or label. |
| Agent | N, a name, or a kind | Agent ordinal, unique agent name, or every initial agent of that kind. |
| Explicit ID | w1:p2, w1:t2, w1 | A pane, tab, or workspace handle from Herdr. |
In agent mode, a pane ID must contain a recognized agent. A kind such as claude expands to matching agents; it does not create an agent.
lictor -p '2:1.1,3:1.2' -s 5
lictor -T -p 'api:2,web:1' -s 5
lictor -A -p 'claude,reviewer' -s 5
Untyped mixed workspace/agent selectors preserve the original precedence: a matching workspace wins before an agent. For mixed selections, prefer explicit typed selectors:
lictor -T -W -A --targets 'workspace=api,tab=w2:t3,agent=reviewer'
The typed prefixes are workspace=, tab=, agent=, and pane=. This avoids confusion between a workspace called claude and the claude agent kind.
Split lanes have separate selectors:
lictor -S w,a --left-targets 'api,web' --right-targets 'reviewer,codex'
lictor -S t,a --left-targets 'w1:t1,w2:t2' --right-targets 'reviewer'
Use lane flags instead of -p in split mode. Each explicit ID must match its lane kind. Repeated entries remain repeated; any nonexistent requested target rejects the whole invocation before view creation.