The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands using variable input in SKILL.md: python3 scripts/resolve_style_guides.py <changed-path>.... This pattern lacks explicit sanitization or escaping instructions, posing a risk of command injection if the file paths contain shell metacharacters.
[DYNAMIC_EXECUTION]: The references/quality-gate-command-matrix.md contains various shell commands (e.g., npm test, cargo test, uv run ruff) that are intended to be executed based on file triggers. Automatic execution of these commands creates a surface for arbitrary code execution if an attacker can manipulate project configuration files like package.json or Cargo.toml.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and review external data in the form of .sql artifacts and other source code files.
Ingestion points: SQL schema, migration, and query artifacts (SKILL.md).
Boundary markers: Absent. There are no instructions for the agent to ignore or delimit instructions potentially embedded within the content of the files it processes.
Capability inventory: The skill has the ability to execute local scripts (scripts/resolve_style_guides.py) and various project-level test/lint commands.
Sanitization: Absent. The skill lacks instructions for sanitizing file content or validating the integrity of the files before processing and execution.

sql-style-guide