The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes the Bash tool to execute a local utility script, scripts/audit-citations.py, to perform citation validation.
[COMMAND_EXECUTION]: The skill uses shell command substitution ($(grep ...)) to dynamically extract the citation style from CLAUDE.md and pass it as an argument to the Python script.
[DATA_EXPOSURE]: The skill reads content from the chapters/ directory and literature/reading_notes/ to identify inconsistencies. This access is restricted to the project workspace and aligns with the skill's stated purpose of auditing thesis drafts.
[INDIRECT_PROMPT_INJECTION]: The skill processes external, potentially untrusted data from thesis chapters and reading notes, which creates a surface for indirect prompt injection.
Ingestion points: Files located in the chapters/ and literature/reading_notes/ directories (SKILL.md).
Boundary markers: None identified; the agent reads and parses these files directly to extract quantitative claims and terminology.
Capability inventory: The skill has access to the Bash tool, allowing it to execute shell commands.
Sanitization: No sanitization or validation of the text extracted from chapters is described before the agent processes it or uses it in the audit report.

audit