sjnims/plugin-dev

AI Summary: The issue proposes adding a new "Shell Script Quality" section to the `CONTRIBUTING.md` file. This section would provide guidance on using `shellcheck` for shell scripts within the plugin, including how to run it, the expected compliance level, and how to disable warnings for intentional patterns. The goal is to ensure consistent quality and best practices for shell scripts.

AI Summary: This issue proposes adding a `skills` frontmatter field to three existing agents (`agent-creator`, `plugin-validator`, `skill-reviewer`). This would enable agents to automatically load relevant knowledge and best practices upon invocation, eliminating the need for users to manually load skills. Specific YAML examples are provided for each agent, detailing which skills should be pre-loaded.

AI Summary: This GitHub issue identifies minor inconsistencies and clarity improvements in a plugin's three subagents when compared to official Claude Code documentation. It proposes three specific changes: aligning the `tools` field format, adding explicit script paths in `plugin-validator`, and improving documentation for `max_tokens` and `temperature` fields. The goal is to enhance alignment with official docs and improve the overall developer experience.

AI Summary: This GitHub issue requests an update to the YAML frontmatter for the `create-plugin` command documentation. It identifies two specific style issues: the `description` field exceeds the recommended 60-character limit, and the `argument-hint` field does not follow the best practice of using square brackets. The goal is to align the frontmatter with Claude Code slash command best practices for better display and consistency.

AI Summary: This GitHub issue requests a documentation fix in the `mcp-integration` skill reference file at line 409. The problem is the use of second-person ("you're") in instructional text, which violates the skill writing style guidelines that mandate imperative/infinitive form. The issue provides the current text, explains why other "you" occurrences are acceptable, and offers two specific options for improvement.

AI Summary: The `mcp-integration` skill's `SKILL.md` documentation is missing a reference to the `search-mcp-servers.sh` utility script in its "Additional Resources" section. This script helps users discover MCP servers with GitHub star rankings. The issue proposes adding a new "Scripts" subsection to `SKILL.md` to improve the discoverability of this useful tool.

AI Summary: The `mcp-integration` skill documentation currently lacks guidance on how to discover relevant MCP servers, despite thoroughly covering server configuration. Users struggle with manual browsing, ineffective search, and blind quality evaluation. The proposed improvement is to add a new `discovery.md` reference file and update the main `SKILL.md` to provide clear guidance on finding suitable MCP servers for plugin needs.

AI Summary: The GitHub issue proposes adding a `search-mcp-servers.sh` script to the `mcp-integration` skill within Claude Code. This script aims to address the current manual and tedious process of discovering MCP servers for plugin development by providing programmatic, contextual search with popularity ranking, thereby eliminating the need for manual browsing and context switching.

AI Summary: This GitHub issue requests a clarification in the `command-development` skill documentation regarding the `model` frontmatter field. Currently, the documentation only lists shorthand values like `haiku` or `sonnet`, but official examples show that full model strings (e.g., `claude-3-5-haiku-20241022`) are also accepted. The suggested change is to update the documentation to explicitly state that both shorthand and full model ID formats are valid.

AI Summary: The issue points out that the `allowed-tools` frontmatter field, which enables restricting tool access for active skills, is documented in the official Claude Code documentation but is missing from the `skill-development` skill's documentation. This omission means users are unaware of a valuable feature for creating read-only or security-sensitive skills. The proposed solution is to add a new section detailing `allowed-tools` to the `SKILL.md` file within the `skill-development` plugin.

AI Summary: This GitHub issue identifies two documentation gaps within the `skill-development/SKILL.md` file. It proposes updating the skill validation checklist to include specific format and length requirements for skill names and descriptions, which are currently missing compared to official Claude Code documentation. Additionally, it aims to clarify the distinct purposes of two reference files listed in the "Additional Resources" section, as their individual roles are not clearly explained.

AI Summary: The `validate-agent.sh` script's name validation regex currently allows uppercase letters, which contradicts the `SKILL.md` documentation stating that agent names should only contain lowercase letters, numbers, and hyphens. This issue proposes updating the regex in the script to enforce the lowercase requirement, aligning the validation with the documented specification.

AI Summary: This GitHub issue addresses a minor style guideline violation in the `agent-development/SKILL.md` file, where a warning emoji (⚠️) is used in a callout. The project's style guidelines state that emojis should be avoided unless explicitly requested. The proposed solution is to replace the emoji with text like "Important - Field Name Difference" or "Warning:" to maintain consistency.

AI Summary: This GitHub issue proposes adding a new utility script, `create-agent-skeleton.sh`, to the `scripts/` directory. This script will generate a template Markdown file for a new agent, including YAML frontmatter, a placeholder description, and a basic system prompt structure, thereby streamlining the agent development process.

AI Summary: This GitHub issue proposes enhancing the discoverability of the `agent-development` skill by adding specific trigger phrases to its description in `SKILL.md`. The current description lacks common comparative and conceptual queries, preventing users from finding the skill when asking questions like "difference between agent and skill." Adding these phrases will improve matching and direct users to the skill's existing helpful comparison table.

AI Summary: The `agent-development` skill's documentation currently provides a comparison of agents, commands, and skills, along with guidance on when to use each. However, it lacks cross-references to the dedicated development skills for commands and skills. The proposal is to enhance the `SKILL.md` by adding explicit links or notes to the `command-development` and `skill-development` skills within the comparison table and relevant sections.

AI Summary: This GitHub issue proposes two documentation enhancements for the agent-development skill. It suggests adding guidance on when to choose specific AI models (haiku, sonnet, opus) and enriching the 'Additional Resources' section with one-line descriptions for reference files.

AI Summary: This GitHub issue proposes adding a new section to the `agent-development` skill documentation. The goal is to clarify for users when to choose between agents, commands, or skills by providing a comparison table. This table will outline the best use cases, triggering mechanisms, and examples for each component.

AI Summary: The `agent-development` skill documentation in `SKILL.md` incorrectly references a non-existent script, `test-agent-trigger.sh`, causing documentation inaccuracy and potential user frustration. The recommended fix is to remove this erroneous reference from the markdown file.

AI Summary: This GitHub issue proposes refactoring content within `SKILL.md` files to adhere to specific writing style guidelines. The goal is to convert passive voice and second-person constructions into imperative or infinitive forms, ensuring a consistent, verb-first instructional style. Examples are provided to illustrate the required conversions, such as changing "Understanding agent structure enables..." to "To create agents effectively, understand the structure...".