The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to process external, untrusted data in the form of reviewer comments. This creates a surface for indirect prompt injection, though the impact is mitigated by the skill's lack of high-risk capabilities.
Ingestion points: User-provided reviewer comments (text or files) as described in SKILL.md.
Boundary markers: The instructions do not define specific delimiters or "ignore embedded instructions" guards for the external comment text.
Capability inventory: The skill is limited to drafting text; it does not utilize network, file-write, or system execution tools.
Sanitization: The skill does not perform sanitization on the input text before processing.
[DATA_EXPOSURE]: The skill includes instructions to read a specific local knowledge file (~/.claude/skills/ml-paper-writing/references/knowledge/paper-miner-writing-memory.md) to improve writing quality. This reference targets a related academic writing skill's memory and does not attempt to access sensitive system files, environment variables, or user credentials.

review-response