The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from project files during the scanning and fixing phases, creating a potential vector for malicious instructions to influence the agent.
Ingestion points: The skill reads all project files during "全项目扫描模式" (Full project scan) and "单文件模式" (Single file mode) as described in SKILL.md.
Boundary markers: No specific boundary markers or delimiters are defined to isolate the content of the analyzed files from the agent's core instructions.
Capability inventory: The skill has the capability to write files (refactoring and splitting files) and execute shell commands (npm test, go test, pytest, and /review quick) as specified in the "修复流程" (Repair process) in SKILL.md.
Sanitization: There is no evidence of sanitization or filtering of the content read from files before it is processed by the agent.
[COMMAND_EXECUTION]: The skill is instructed to identify and run test commands found within the project environment.
Evidence: SKILL.md states: "检测项目是否有测试命令（npm test / go test ./... / pytest 等）" and "运行测试". This involves executing arbitrary commands defined in project configuration files (e.g., package.json).
[DYNAMIC_EXECUTION]: The skill performs dynamic execution by running the project's own tests to verify refactoring results.
Evidence: The verification step in SKILL.md involves executing local code via test runners, which is a form of dynamic execution of local project assets.

cc-size-check