The Agent Skills Directory

[COMMAND_EXECUTION]: The skill relies on local command execution to perform its primary functions. scripts/generate-report.js uses child_process.execFileSync to invoke the GitHub CLI (gh). Additionally, scripts/html-to-pdf.sh executes system binaries such as google-chrome, chromium, and wkhtmltopdf to generate report artifacts.
[EXTERNAL_DOWNLOADS]: The skill has dependencies on external tools and packages. scripts/puppeteer-pdf.js requires the puppeteer Node.js package. The shell script scripts/html-to-pdf.sh assumes the presence of external tools like chrome or wkhtmltopdf for PDF conversion.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data which could contain malicious instructions.
Ingestion points: Pull request titles, descriptions, and labels are retrieved from external GitHub repositories using the gh CLI in scripts/generate-report.js.
Boundary markers: The narrative generation instructions in SKILL.md and reference/retrospective-voice.md do not implement explicit boundary markers or instructions to ignore embedded commands in the PR content.
Capability inventory: The skill possesses the ability to execute shell commands and perform file system operations.
Sanitization: While scripts/generate-report.js includes an escapeHtml function for the final report display, the agent's internal reasoning and narrative aggregation phases do not specify sanitization for the PR data it analyzes.
[DYNAMIC_EXECUTION]: The skill involves the runtime generation and execution of code and scripts. It generates HTML reports dynamically and then invokes external browser engines or Node.js scripts (scripts/puppeteer-pdf.js) to process these generated files into PDF format.

harvest