-spec affected_siblings(TargetPid :: pid()) -> {ok, [pid()]} | {error, term()}.

Return the PIDs of siblings that would be killed or restarted if TargetPid dies.

Depends on the parent supervisor's strategy:

• one_for_one: TargetPid only
• one_for_all: all siblings
• rest_for_one: TargetPid and all siblings after it
• simple_one_for_one: TargetPid only

-spec kill_graph(TargetPid :: pid()) -> {ok, [pid()]} | {error, term()}.

Return all PIDs that could cause TargetPid to be killed.

This is the set of ancestors unioned with siblings that trigger cascade restarts under the parent supervisor's strategy:

• one_for_one / simple_one_for_one: only ancestors (no sibling killers)
• one_for_all: all siblings are killers
• rest_for_one: siblings at positions before the target are killers

-spec kill_graph_detail(TargetPid :: pid()) -> {ok, [ides_family:kill_source()]} | {error, term()}.

Return the kill graph with each entry tagged by its kill mechanism:

• ancestor — supervisor ancestor
• sibling — sibling via supervisor strategy
• link — linked process (relevant if target doesn't trap exits)
• monitor — monitored process (relevant if target doesn't handle DOWN)

-spec link_info(Pid :: pid()) -> {ok, ides_family:link_info()} | {error, term()}.

Return link information for the given process.

Reports which processes are linked to Pid and whether Pid traps exits. Linked processes are potential killers if traps_exits is false.

-spec monitor_info(Pid :: pid()) -> {ok, ides_family:monitor_info()} | {error, term()}.

Return monitor information for the given process.

Reports which processes Pid is monitoring and which processes are monitoring Pid. Monitored processes are potential killers if Pid doesn't handle DOWN messages.

-spec should_restart(Pid :: pid(), Reason :: exit_reason()) -> boolean().

Return whether a terminated child Pid would be restarted by its supervisor.

Rules:

• permanent: always restarted
• transient: restarted only on abnormal exit
• temporary: never restarted