CommandCodeAI / CommandCodeAI/command-code
Bug: /design skill hangs on longer prompts regardless of flags
@naymurdev ya está trabajando en esto.
Desde el 29/6/2026.
- Lenguaje dominante
- Sin datos de lenguaje
- Estrellas
- 4k
- Forks
- 350
- Métricas de merge de PR
- Sin PR fusionados en 30 d
Descripción
/design skill hangs on longer prompts regardless of flags
Environment:
- Command Code v0.27.0
- Model: default (taste-1)
- Platform: Linux (x64), Node v24.15.0
- Host: local machine
Steps to reproduce:
- Run
cmdwith the/designskill - Submit a comprehensive design review prompt (covering UI audit of a multi-page application)
- Command Code hangs indefinitely — no progress, no output, no error
- Must be killed externally (SIGKILL)
Observed behavior:
/designworks reliably with--max-turns 10and a short prompt (2-3 sentences)- With longer prompts (describing a full site review) or without
--max-turnslimit, the process freezes after the first response - The hang is not a timeout — it persists for hours with no CPU activity
- Occurs regardless of
--max-turns,-p(non-interactive), or interactive flags - No error logs are written; no crash report
Expected behavior:
/designshould process prompts of any reasonable length- Should handle comprehensive design reviews spanning multiple pages/sections
- Should either complete, provide partial output, or return a clear error
Impact:
Blocks using /design for meaningful site-wide design audits. Only trivial single-component prompts work.
Additional context:
This was discovered while trying to run a full-site design audit of a React 19 application. The same review was successfully completed using alternative tools (impeccable detector + manual UI audit scripts), confirming the prompt content itself is valid.
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.
Evaluación
Este issue todavía no se ha evaluado.