The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from the local codebase and repository metadata to generate an HTML report, creating a surface for cross-site scripting (XSS) or data-driven instruction override.
Ingestion points: The skill reads file names, code contents, git log output, and existing ADRs (Architectural Decision Records) from the repository being scanned.
Boundary markers: No explicit boundary markers or instruction-shielding delimiters are defined for the data interpolated into the HTML template.
Capability inventory: The skill has the capability to write files to the system's temporary directory, execute shell commands (git, browser launchers), and generate complex diagrams via Mermaid.js.
Sanitization: The skill's HTML template explicitly sets Mermaid's securityLevel to "loose". This configuration permits certain HTML tags within diagrams, which could be exploited if a malicious file name or code comment in the analyzed repository is rendered into a diagram, leading to script execution in the user's browser when the report is opened.
[COMMAND_EXECUTION]: The skill utilizes several system commands to perform its functions. It executes git log --oneline to identify codebase 'hot spots' and uses platform-specific commands (xdg-open, open, start) to automatically launch the generated architectural report in the default web browser. These operations are consistent with the skill's documented behavior for history analysis and user presentation.
[EXTERNAL_DOWNLOADS]: The generated HTML report includes script tags that fetch the Tailwind CSS framework and the Mermaid.js visualization library from reputable content delivery networks (tailwindcss.com and jsdelivr.net). These external resources are standard tools for modern web-based visual reporting and are used here for layout and diagram rendering.

improve-codebase-architecture