The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses bun run to execute local TypeScript scripts provided within the skill package (prepare-humanize-input.ts and verify-change-rate.ts). These scripts perform text processing, sanitization, and metrics calculation without external network dependencies.
[INDIRECT_PROMPT_INJECTION]: The skill processes user-provided text to remove 'AI-like' patterns. It implements robust safety guardrails, including a sanitization script that removes invisible characters and chatbot boilerplate, as well as strict instructions to the agent (e.g., 'Fidelity First', 'Do-NOT' list) to prevent the alteration of facts, numbers, or direct quotes.
[SAFE]: The skill instructions explicitly restrict the agent's file system activities to relative paths and a designated workspace directory, forbidding the use of discovery commands like glob or grep to explore the host environment.

lang-lgnh