The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes a suite of environment hooks (PreToolUse, PostToolUse, and Stop) to execute local shell scripts such as multi_linter.sh and protect_linter_configs.sh whenever file modifications occur.
[DYNAMIC_EXECUTION]: Implements a three-phase architecture that spawns claude -p subprocesses to automatically generate and apply code fixes at runtime based on linter violations.
[INDIRECT_PROMPT_INJECTION]: The automated repair pipeline creates an attack surface where untrusted data (project source code) is processed by linters and then fed into a sub-agent for remediation.
Ingestion points: Any project files being edited or created (Python, TypeScript, Shell, YAML, etc.).
Boundary markers: The documentation does not specify explicit boundary markers or 'ignore' instructions for the sub-agent prompts.
Capability inventory: The system executes shell scripts via hooks and performs file-write operations through the claude -p sub-agent.
Sanitization: Linter output is structured as JSON before being passed to the sub-agent, though the sub-agent must still process the original code context to perform fixes.
[EXTERNAL_DOWNLOADS]: The documentation instructs users to install various external CLI tools and linters (e.g., ruff, uv, biome, hadolint, jaq) using system package managers like Homebrew.

plankton-code-quality