The Agent Skills Directory

[PROMPT_INJECTION]: The skill workflow involves processing user-provided file content and feedback, creating a surface for indirect prompt injection.
Ingestion points: The skill reads local SKILL.md files and collects user responses through AskUserQuestion calls as part of the lesson blocks.
Boundary markers: Content from the user-provided SKILL.md files is not wrapped in specific boundary markers or delimiters during ingestion.
Capability inventory: The agent performs interactive questioning and suggests reading and editing local files through the command-line interface.
Sanitization: No sanitization or safety-specific validation of the ingested SKILL.md content is performed before processing.
[EXTERNAL_DOWNLOADS]: The skill references resources from a trusted organization.
Evidence: Includes links to official documentation on Anthropic's domain (docs.anthropic.com) for learning about skills and multi-agent systems.
[NO_CODE]: The skill consists entirely of instructional markdown files and does not include any executable scripts or binary files.

session4-polish