The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes a local Python module using PYTHONPATH=src python3 -m siyu_team.eval.cli to perform compliance scoring on generated reports. This is a core functional component of the skill's workflow.
[INDIRECT_PROMPT_INJECTION]: The skill processes content from multiple local archives to generate a summarized report, which presents a surface for indirect prompt injection if those archives contain untrusted instructions.
Ingestion points: Data is read from files within ~/.siyu/clients/ (SKILL.md).
Boundary markers: The skill expects a specific structure (frontmatter and six sections) but does not define explicit delimiters to isolate archive content from instructions.
Capability inventory: The skill has the ability to read/write files and execute local Python code via shell commands.
Sanitization: A mandatory compliance scan (siyu_team.eval.cli) is performed to detect and mitigate 'COMPLIANCE_RED' content.
[DATA_EXPOSURE]: The skill reads client data from local directories and writes consolidated reports. It includes a specific security constraint to '禁止路径穿越' (forbid path traversal) to ensure file operations remain within the designated client folders.

siyu-report