The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests data from external job descriptions and user-supplied past documents, which creates a surface for indirect instructions to influence agent behavior.
Ingestion points: Processes files at ~/career-applications/[job-slug]/01-job-analysis.md and reads content from user-provided cover letters.
Boundary markers: No explicit delimiters or instructions to disregard embedded commands are included in the instructions for processing these files.
Capability inventory: The skill is designed to write new files to the filesystem (04-cover-letter-framework.md and .json export files).
Sanitization: There are no documented sanitization or filtering steps for the ingested external content.
[DATA_EXFILTRATION]: The skill is configured to access sensitive personal career directories.
Sensitive paths: Accesses ~/lexicons_llm/ and ~/career-applications/ to retrieve career philosophy and application history.
Risk assessment: This file access is essential to the skill's stated purpose of personalizing cover letters. The risk is mitigated by the absence of network-capable tools or outbound requests, ensuring data remains on the local system.

cover-letter-voice