plankton-code-quality

Pass

Audited by Gen Agent Trust Hub on Jun 24, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill utilizes Claude Code hooks (PreToolUse, PostToolUse, and Stop hooks) to execute shell scripts such as multi_linter.sh and protect_linter_configs.sh. These scripts are used to enforce code quality, manage package installations (blocking pip/npm in favor of uv/bun), and prevent the agent from modifying linter configurations to bypass rules.
  • [PROMPT_INJECTION]: The skill presents an indirect prompt injection surface (Category 8). It processes data from edited files and linter reports, which are then fed into automated remediation subprocesses using the claude -p command.
  • Ingestion points: Local source code files being edited and structured JSON violation data (Phase 2 and 3).
  • Boundary markers: The system uses structured JSON to pass violation context to the subprocess.
  • Capability inventory: Execution of shell hooks and spawning of additional LLM subprocesses for code modification.
  • Sanitization: No specific sanitization or filtering of code comments or content is described before passing data to the remediation model.
  • [EXTERNAL_DOWNLOADS]: The setup instructions require the installation of various well-known external tools and linters, such as jaq, ruff, uv, biome, shellcheck, and hadolint, which are managed via system package managers like Homebrew or NPM.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 24, 2026, 12:58 PM
Security Audit — agent-trust-hub — plankton-code-quality