The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses local command-line tools including git, grep, and glob, as well as a project-specific script python .claude/scripts/code_graph. These are standard tools for identifying and analyzing code patterns within a repository.
[PROMPT_INJECTION]: The instructions utilize emphatic and repetitive language such as "MANDATORY IMPORTANT MUST ATTENTION" and "NON-NEGOTIABLE" to enforce its quality control protocols. While these keywords are often found in instruction override attempts, they are used here to ensure the agent adheres to the skill's specific logical workflow.
[PROMPT_INJECTION]: The skill exhibits an inherent surface for indirect prompt injection as it processes untrusted source code and documentation.
Ingestion points: Code files identified through git diff or user-provided paths, and project-specific documentation files (SKILL.md).
Boundary markers: No explicit delimiters are employed to isolate the content of files being read from the agent's instructions.
Capability inventory: The agent can modify files using Edit and execute shell commands via Bash (SKILL.md).
Sanitization: There is no evidence of sanitization or filtering applied to the ingested code before the agent performs analysis or refactoring.

code-simplifier