Expand description
Engine-level command results — TUI-agnostic outcomes of slash commands.
The engine processes a command and returns a CommandResult.
Renderers (TUI, headless) decide how to display the result.
Structs§
- Session
Summary - Summary of a session for listing.
Enums§
- Command
Result - Result of processing a slash command in the engine.
- Modal
Request - TUI-specific modals the engine can request.
- Thinking
Spec - A thinking specification — either a named level or a custom numeric budget.
Functions§
- budget_
command - Explicit operator control of future turns. No config writes, inference, history mutation, or changes to an already-running stream’s budget snapshot. This is a frontend slash command, never a model-callable budget renewal.
- context_
command /context(Task 12, spec §6.6 acceptance): structured metadata-only report — system/tool/history counts and bytes, latest cache component change and reuse estimates, writer counters. Surfaces that own the conversation pass it ashistory; others get honestunavailable.- evaluate_
engine_ command - Pure command → result mapping (no runtime mutation).
- handle_
engine_ command - Process commands that are pure engine logic — no TUI state needed. Returns None if the command needs TUI-level handling.
- memory_
command /memory on|recall|capture|once|status|off|index-history|why(task A5, spec §7.3): the ONE deterministic engine entry point every frontend — TUI, headless chat, RPC, server, watcher, agent — shares viahandle_engine_command. Frontends must not duplicate lease or budget logic.- parse_
command - Parse a slash command into (command, arg).
- parse_
thinking_ arg - Parse a
/thinkingargument into aThinkingSpec. - persist_
to_ config - Persist a config key and return a user-visible status suffix.
- thinking_
config_ value - Canonical config-file string for a thinking change. Returns the exact budget digits for Custom, named string for Named.
- trace_
command /trace next|next content|status(Task 12): explicit trace controls.