The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes a wide range of shell commands to synchronize the repository (git pull --rebase, git stash), manage branches (git checkout -b), and run various static analysis tools dependent on the detected project stack.
[EXTERNAL_DOWNLOADS]: The orchestrator is instructed to verify and potentially run tools via npx, which can lead to the download and execution of arbitrary packages from the npm registry at runtime.
[INDIRECT_PROMPT_INJECTION]: The skill has a significant attack surface for indirect prompt injection because it processes untrusted content from the target codebase.
Ingestion points: The skill reads all files within the repository for analysis by eight specialized subagents.
Boundary markers: The instructions do not define clear boundaries or 'ignore' instructions for content found within the codebase being cleaned.
Capability inventory: The skill can modify and delete files, commit changes to git, and execute arbitrary shell commands for testing and typechecking.
Sanitization: There is no evidence of sanitization or filtering of the code content before it is processed by the agent's reasoning engine.
[DYNAMIC_EXECUTION]: The skill orchestrates eight specialized subagents using a parallel execution pattern, where each agent is given a specific role and prompt to perform deep cleanup tasks.

slop-code