application-cleanup-method
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill resolves and executes
BUILD_CMDandTEST_CMDfrom project configuration files such aspackage.jsonandpyproject.toml. These commands are executed during the baseline phase and as gates for each cleanup phase. - [INDIRECT_PROMPT_INJECTION]: The skill operates on untrusted data from the target project's codebase to drive cleanup decisions and validation, creating a surface for indirect prompt injection.
- Ingestion points: Target project source code, documentation files (
*.md,*.mdx,*.rst,*.adoc), and configuration manifests (package.json,pyproject.toml,CLAUDE.md). - Boundary markers: None explicitly defined to isolate untrusted codebase content from the agent's instructions during processing.
- Capability inventory: File system modifications (file deletion,
git rm, line-level edits), Git state management (snapshots, resets), and execution of arbitrary project-defined scripts for building and testing. - Sanitization: Employs Grep-based reference checking and build/test success gates to validate deletions, though these measures do not sanitize the content of the files being read or the commands being executed.
Audit Metadata