The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes subprocess.Popen in scripts/run_eval.py to invoke the claude CLI for automated trigger testing. The eval-viewer/generate_review.py script also uses subprocess.run to execute lsof and kill commands for local port management during the serving of review pages.- [DYNAMIC_EXECUTION]: The toolkit dynamically constructs temporary markdown files in the project's .claude/commands/ directory to simulate skill environments during evaluation cycles. It also performs runtime rendering of spreadsheet data in the browser using the SheetJS library.- [INDIRECT_PROMPT_INJECTION]: The workflow ingests user-provided business context and test prompts which are then processed by sub-agents for grading and analysis. The skill mitigates risks associated with untrusted data by implementing a mandatory '6-question gate' that enforces specific validation and quality standards before data is accepted into the agent's context.- [EXTERNAL_DOWNLOADS]: The eval-viewer/viewer.html component references the SheetJS library from an external CDN (cdn.sheetjs.com) to enable spreadsheet visualization for the user during the evaluation review process.

zach-seller-skill-creator