The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze potentially untrusted external data to customize agent behavior. * Ingestion points: The skill processes active/archived sessions, exported transcripts, shell history, and community-created skill files as described in references/audit-rubric.md and references/personalization-rubric.md. * Boundary markers: No explicit technical delimiters or instructions are provided to the agent to isolate embedded instructions in the analyzed data from the core auditing logic. * Capability inventory: The skill allows for the inspection and direct editing of SKILL.md files and associated resources within the agent's environment. * Sanitization: Although references/audit-rubric.md suggests using sanitized examples in reports, there is no technical enforcement or validation mentioned for the data analysis process itself.

skill-personalizer