The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes shell commands (git diff, ruff check, uv run mypy, and pytest) to perform its core functions. This creates a potential command injection surface if a malicious actor creates files with names containing shell metacharacters (e.g., ; rm -rf / ;.py), which might be executed if the agent's shell tool does not adequately sanitize inputs.
[PROMPT_INJECTION]: The skill reads and processes local source code, passing it to external tools and a sub-agent (code-simplifier:code-simplifier). This allows for indirect prompt injection, where malicious instructions embedded in code comments or string literals within the analyzed files could attempt to manipulate the agent's logic or the sub-agent's output.
Ingestion points: Local Python files identified in the synapse/ or tests/ directories via git diff or directory scanning.
Boundary markers: None. The skill does not specify the use of delimiters or 'ignore instructions' markers when passing code content to tools or sub-agents.
Capability inventory: Executes shell commands and invokes sub-agents based on file content.
Sanitization: The skill only filters for the .py file extension and does not sanitize or validate the content of the files before processing.

code-quality