The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user-provided manuscripts through various automated functions and shell tools. This creates an attack surface where maliciously crafted input text could attempt to influence the agent's behavior during the polishing process.
Ingestion points: User-supplied text is ingested by functions like identify_ai_patterns in ai-writing-humanizer/SKILL.md and check_text_with_languagetool in grammar-checker-guide/SKILL.md, and processed via file-based shell commands in academic-writing-refiner/SKILL.md.
Boundary markers: The implementation does not include explicit boundary markers or instructions for the agent to ignore instructions embedded within the processed text.
Capability inventory: The skill possesses capabilities for network operations via urllib.request in grammar-checker-guide/SKILL.md and shell command execution including java -jar, npm install, chktex, and aspell across multiple files.
Sanitization: There is no evidence of input validation or sanitization performed on the manuscript text before it is processed or sent to external APIs.
[EXTERNAL_DOWNLOADS]: The skill instructs the agent to install external packages to facilitate text linting.
Evidence: Instructions in academic-writing-refiner/SKILL.md command the global installation of textlint and associated rules from the NPM registry.
[COMMAND_EXECUTION]: The skill provides instructions and scripts that trigger shell command execution on the host system to perform academic polishing tasks.
Evidence: The agent is directed to use java -jar for LanguageTool CLI, npm install for linter setup, and technical tools such as chktex, aspell, and grep for manuscript analysis in paper-polish-guide/SKILL.md and academic-writing-refiner/SKILL.md.

polish-skills