Feature request: `/btw` side-channel for asking questions without interrupting an active run
Nadie ha tomado este issue todavía.
Evaluación
- Dificultad
- 5/5
- Tiempo estimado
- Más de una semana
- Aptitud para principiantes
- 35/100
- Tipo de issue
- Nueva funcionalidad
- Claridad
- Necesita aclaración
- Estado de actividad
- Activo
- Área
- cli
Línea de trabajo
Start by locating the TUI input path and the state exposed for the active foreground run, including task/todo progress and background-agent status. Compare those surfaces with the existing Background panel and determine how a non-steering query could read an observational snapshot. Done means a user can ask about the active run without interrupting, queueing, cancelling, or changing its context.
Escrito por el modelo de indexación a partir del texto del issue.
Descripción
Feature Description
Add a lightweight side-channel (for example /btw, a small secondary prompt, or an equivalent TUI panel) that lets the user ask a question while the main agent is actively running, without interrupting, steering, cancelling, or queueing a new turn into that run.
The key requirement is that this side conversation should be observational and non-destructive: the foreground task keeps running exactly as-is, while the user can ask things such as:
- "What step are you on right now?"
- "Which year/file are you processing?"
- "What remains after the current step?"
- "Are the background/sub-agents still running?"
- "What was the last completed action?"
A response could be based on the current task/todo state, recent tool activity, active sub-agents/background jobs, and any already-written files. It should not inject the side question into the main model context unless the user explicitly chooses to do so.
Use Case
For long autonomous tasks, the current UX makes it hard to check progress without disturbing execution. For example, if Command Code is processing a large multi-year dataset or editing many files, I may want to know whether it has reached 2018, 2020, etc. while letting the main run continue uninterrupted.
Today the closest workarounds are:
- Open a second Command Code session in the same project and ask it to inspect the filesystem read-only.
- Use the Background panel for shell tasks.
- Poll sub-agent status where available.
These are useful but not equivalent. A second session has no direct view of the foreground run's internal task/todo state, and the Background panel is primarily about background work rather than the active foreground agent.
A /btw-style side-channel would make long-running sessions much easier to supervise without changing their behavior.
Possible UX
One possible design:
/btw what step are you on?
or a keyboard shortcut that opens a small temporary prompt/panel over the TUI.
The answer could show a compact snapshot such as:
Main run: active
Current task: Processing 2019 exam solutions
Last completed: 2018.md + 6 cropped images
Remaining: 2020-2023
Background agents: 2 running
Important semantics:
- Does not interrupt or cancel the active run.
- Does not become a steering message by default.
- Does not alter the active run's context/todos.
- Can inspect current run state and already-observable project state.
- Optional action to "send this to the main run" if the user actually wants to steer it.
Additional Context
There are related issues around background-agent visibility (for example #874), but this request is specifically about a side conversation/status query for an active foreground run, rather than improving the telemetry of detached sub-agents.
This pattern is especially useful for long autonomous coding/data-processing tasks where users want supervision without interference.
- Lenguaje dominante
- Sin datos de lenguaje
- Estrellas
- 4k
- Forks
- 350
- Métricas de merge de PR
- Sin PR fusionados en 30 d
Guía de contribución
No hay ninguna guía de contribución indexada para este repositorio
Primeros pasos
- Lee el issue completo y luego la guía de contribución del proyecto.
- Comenta en el issue que vas a ocuparte — evita que dos personas hagan lo mismo.
- Haz un fork del repositorio y trabaja en una rama.
- Abre un pull request que haga referencia al número del issue.
Más de CommandCodeAI/command-code
-
Dificultad 2/5 1-3 horas Aptitud para principiantes 68/100
CommandCodeAI/command-code#903 · 2 comentarios ·
-
Dificultad 2/5 1-3 horas Aptitud para principiantes 68/100
CommandCodeAI/command-code#855 ·
-
Dificultad 2/5 1-3 horas Aptitud para principiantes 78/100
CommandCodeAI/command-code#841 · 1 comentario ·
-
Dificultad 2/5 1-3 horas Aptitud para principiantes 68/100
CommandCodeAI/command-code#655 · 1 comentario ·
-
Dificultad 2/5 1-3 horas Aptitud para principiantes 68/100
CommandCodeAI/command-code#608 ·
Todos los issues de CommandCodeAI/command-code
Issues similares
-
triage
Dificultad 1/5 1-3 horas Aptitud para principiantes 88/100
-
Dificultad 2/5 1-3 horas Aptitud para principiantes 68/100
-
app bug
Dificultad 2/5 1-3 horas Aptitud para principiantes 72/100
-
Dificultad 2/5 1-3 horas Aptitud para principiantes 73/100
-
Broadcast Documentation Abierto
Dificultad 1/5 Menos de una hora Aptitud para principiantes 90/100
kovidgoyal/kitty#10516 ·