The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill clones the vendor's official documentation repository (github.com/ZenMux/zenmux-doc) to retrieve the latest GitHub issue templates.
[COMMAND_EXECUTION]: The skill utilizes several system commands to perform its tasks:
Executes git to clone and pull updates for reference materials.
Uses the GitHub CLI (gh) to verify authentication status and create issues on the remote repository.
Runs local shell scripts (scripts/update-references.sh, scripts/get-doc-tree.sh) for environment setup and documentation indexing.
[DYNAMIC_EXECUTION]: The scripts/update-references.sh file uses a python3 one-liner to dynamically calculate relative paths between the project root and the references directory. This is used for correctly updating the project's .gitignore file.
[INDIRECT_PROMPT_INJECTION]: The skill acts as a conduit for user-provided data and external templates, creating a vulnerability surface for indirect prompt injection.
Ingestion points: User chat input containing feedback/bug descriptions and .yml template files from the ZenMux/zenmux-doc repository.
Boundary markers: Absent. The skill does not use specific delimiters or warnings to prevent the model from following instructions embedded in the feedback content.
Capability inventory: File system access (reading templates, writing to /tmp, modifying .gitignore) and network access (GitHub API via gh).
Sanitization: The skill employs a secure practice of writing the issue body to a temporary file (/tmp/zenmux-issue-body.md) and passing it to the CLI via a file flag, preventing shell injection from user-controlled strings.

zenmux-feedback