The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute multiple shell commands, including cp to deploy templates and ls to check for overlay directories via environment variables.
[INDIRECT_PROMPT_INJECTION]: The skill has a significant attack surface for indirect prompt injection.
Ingestion points: The skill reads sitemap.xml, local *.html files (extracting titles, meta descriptions, and first paragraphs), and the file aieo/llms-exempt.txt.
Boundary markers: There are no boundary markers or explicit instructions provided to the agent to disregard potential instructions embedded within the titles or paragraphs of the HTML files it processes.
Capability inventory: The agent has the capability to write to the file system (llms.txt), execute shell commands (cp, ls), and run Python scripts.
Sanitization: There is no evidence of sanitization or validation of the external content before it is processed and interpolated into the llms.txt file.
[DYNAMIC_EXECUTION]: In Step 5, the skill generates a Python script at runtime and executes it using a shell here-doc (`python3
... <<'EOF'`). This script performs logic to verify synchronization between the sitemap and the index, representing dynamic script generation and execution.

llms-txt