Description
In some SKILL.md(~20, e.g. architecture_review), there is 'agent: xxx' defined in YAML Schema. But this schema will be ignored if there is no schema 'context: fork' defined at the same time. This rule can be seen in the Claude Code Doc.
Then when we use the slash command, actually the skill won't be handled by specific agent. The main agent will handle this skill.
When I try adding the 'context: fork' schema, I found anthoer conflict stuff:
(Still take architecture_review as example)
- In SKILL.md, 'askUserQuestion' is allowed in tools schema, and in the steps description, 'askUseQuestion' will be used several times.
- In technical_director.md, 'askUserQuestion' is not allowed in tools schema. So /architecture_reviw invodes technical_director as SubAgent to handle the skill, seems it doesn't have permission to ask user questions for clarification.
I am very confuse on this issue now, and very curious about which method is better.
I sincerely hope to receive your support and answers.
Many thanks!
Description
In some SKILL.md(~20, e.g. architecture_review), there is 'agent: xxx' defined in YAML Schema. But this schema will be ignored if there is no schema 'context: fork' defined at the same time. This rule can be seen in the Claude Code Doc.
Then when we use the slash command, actually the skill won't be handled by specific agent. The main agent will handle this skill.
When I try adding the 'context: fork' schema, I found anthoer conflict stuff:
(Still take architecture_review as example)
I am very confuse on this issue now, and very curious about which method is better.
I sincerely hope to receive your support and answers.
Many thanks!