remove-deadcode

Pass

Audited by Gen Agent Trust Hub on Oct 16, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill analyzes codebase content to identify candidates for deletion, which involves reading untrusted files.\n- Ingestion points: The agent reads project source files, export definitions, and symbol references during the Scan and Verify phases.\n- Boundary markers: The instructions do not specify explicit delimiters or warnings to ignore instructions embedded in the code being analyzed.\n- Capability inventory: The agent is instructed to run compiler/linter commands (e.g., tsc, eslint, vulture, cargo), perform file deletions, and execute Git commands for staging.\n- Sanitization: There is no mention of sanitizing or escaping the content read from files before it is processed by the agent.\n- Mitigation: The skill mandates that removals must be verified against barrel exports and dynamic references, and requires a successful build and test pass before the task is considered complete.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 16, 2026, 03:21 AM