housekeep-dead-code

Warn

Audited by Gen Agent Trust Hub on Oct 7, 2026

Risk Level: MEDIUMREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
  • [REMOTE_CODE_EXECUTION]: The skill instructs the agent to execute the remove-unused-vars package via npx (e.g., npx eslint ... | npx remove-unused-vars). This package is explicitly described in the documentation as "highly experimental" (version 0.0.x) and is run without version pinning, creating a risk of executing unverified or malicious code from a public registry.
  • [COMMAND_EXECUTION]: Core functionality relies on extensive shell command execution to perform repository audits and modifications. This includes using npx knip --fix-type for file deletion, jscpd for duplication analysis, and various supabase CLI commands for database schema inspection and migration management.
  • [INDIRECT_PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection via the processing of untrusted data from an external plan file.
  • Ingestion points: The plan-dead-code.md file, which contains the critical keep/kill list used to drive deletion operations (SKILL.md).
  • Boundary markers: None identified; the skill instructions direct the agent to follow the plan's list strictly.
  • Capability inventory: File system deletion (knip), database schema modification (supabase db), and automated code editing (remove-unused-vars).
  • Sanitization: There is no validation or escaping of the plan file's contents before they are used to influence the agent's actions.
  • [DYNAMIC_EXECUTION]: The skill generates and executes code modifications at runtime by piping linter output into the remove-unused-vars tool. This dynamic process modifies the source code based on live tool reports without intermediate human review between the tool's execution and the resulting file changes.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Oct 7, 2026, 11:27 PM