The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to read a 'pqw-skills 接入' section from a project's AGENTS.md file to configure its behavior and command set. This ingestion point processes untrusted project data that could potentially contain malicious instructions. * Ingestion points: Project-specific AGENTS.md file (specifically the 'pqw-skills 接入' section). * Boundary markers: No explicit delimiters or 'ignore' instructions are defined for the project-provided configuration data. * Capability inventory: The skill references external tools and other skills (e.g., slice-closeout, formal-docs-landing, clear-writing) that execute shell commands for linting, testing, and document validation. These capabilities could be misused if the configuration is poisoned. * Sanitization: No specific sanitization or validation logic for the input from AGENTS.md is described in the provided instructional files.
[NO_CODE]: The provided files consist entirely of Markdown documentation and do not contain any scripts, executables, or direct shell command definitions. All referenced logic and tools are external to this specific skill package.

teach-pqw-skills