The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes local system commands including git for change detection, grep for extracting import patterns, and node/python3/uv for running analysis logic.
[COMMAND_EXECUTION]: It dynamically generates a Python script at /tmp/cycle_detect.py to perform circular dependency detection on the extracted module graph data.
[EXTERNAL_DOWNLOADS]: Recommends the installation and use of the graphifyy utility via uv tool install or pipx install for enhanced codebase indexing.
[SAFE]: Utilizes a local MCP tool (mcp__plugin_dh_backlog__artifact_register) to save analysis reports as artifacts, keeping the data within the user's project environment.
[PROMPT_INJECTION]: Ingests untrusted project files and configuration data (e.g., pyproject.toml, source code) as part of the analysis process, presenting an indirect injection surface.
Ingestion points: Uses Glob and Grep to read source code files and project metadata across the repository.
Boundary markers: Does not utilize explicit boundary markers or "ignore instruction" warnings for data parsed from the codebase.
Capability inventory: Includes file system access, command execution, and artifact registration.
Sanitization: Does not implement specific sanitization of the content extracted from third-party or project files before processing.

code-review-architecture