Expand description
Pure outcome -> action policy table (D-08/D-11/D-12, 17-01).
decide_action is the single exhaustive policy surface advance()
(Plan 04) dispatches on. It has no I/O, no CliError, no filesystem, and
no process spawn — deterministic pure function of (Stage, AgentStatus).
The match has NO wildcard arm: adding a future crate::agent_result::AgentStatus
variant without extending this match is a compile error, which is the
mechanism that prevents the D-01 regression class (a new/unhandled
outcome silently advancing).
Enums§
- Action
- The action to take in response to an agent outcome at a given stage.
Functions§
- decide_
action - Decide what to do given the outcome of a stage’s agent run.