Feat/ask questions skill

[fdematos]

No description provided.

[opencode-agent]

Here's my review of the pull request:

Summary

This PR adds a new "ask-questions-if-underspecified" skill to the opencode-froggy plugin, replacing the previous "code-release" skill. The changes are well-documented and follow the existing patterns in the codebase.

Code Quality Issues

1. CHANGELOG.md: The entry mentions "Replace code-release skill with ask-questions skill" but doesn't clearly indicate what happened to the code-release functionality. This could be confusing for users who were using that skill.

2. README.md: The new skill is well-documented, but the "Available Skills" table could benefit from a more detailed description of when to use this skill.

3. skill/ask-questions-if-underspecified/SKILL.md: The skill documentation is comprehensive and well-structured. However, the use_when condition is quite long and might benefit from being more concise.

Potential Bugs

1. src/tools/skill.ts: The skill discovery logic has been updated to include the new skill, but I notice that the skill loading mechanism relies on file system operations that could potentially fail silently (lines 58-60, 63-65). This could make debugging difficult if a skill fails to load.

2. src/index.ts: The skill activation block is built correctly, but there's no error handling if the skill loading fails during plugin initialization.

Suggestions for Improvement

1. Error Handling: Add more robust error handling and logging when skills fail to load, especially in the discoverSkillsFromDir function.

2. Documentation: Consider adding examples of when this skill would be particularly useful in the README.md file.

3. Testing: Ensure there are tests for the new skill, especially for the automatic activation mechanism.

4. Consistency: The skill name in the filesystem is "ask-questions-if-underspecified" but in the documentation it's sometimes referred to as "ask-questions". Consider standardizing on one naming convention.

5. Performance: The skill discovery process scans multiple directories. Consider adding caching or making this process more efficient, especially for large projects.

The changes are generally well-implemented and follow the existing patterns in the codebase. The new skill appears to be a valuable addition that will help clarify ambiguous requirements before implementation begins.

opencode session  |  github run