Use when an agent needs current advisory guidance for HTML, CSS, browser APIs, accessibility, performance, forms, privacy, security, client-side JavaScript, or modern web platform alternatives.
Use when implementing, reviewing, refactoring, simplifying, debugging, or planning code changes that need explicit assumptions, minimal scope, surgical edits, and verifiable success criteria.
Use when an agent needs to choose the correct coding-ethos MCP, cerun, managed lint, code-intel, or remediation path before acting.
Use when an agent needs to run lint, formatting, type checks, or policy-backed diagnostics. Prefer the MCP managed_lint tool, then the ./bin/lint wrapper, instead of invoking raw Ruff, mypy, uvx, host linters, or package-manager shortcuts.
Use when captured linters, generated tool configs, managed binaries, config drift, missing tools, package-relative path resolution, or host tool usage causes lint or hook failures.
Use when git commands, commit hooks, admin-approved commits, protected hook paths, raw git bypass attempts, or commit attribution failures appear in agent or developer workflows.
Use when Ruff, mypy, pyright, pylint, Bandit, SQLFluff, Tombi, or another managed checker reports findings that need structural code-quality remediation rather than suppressions or config weakening.
Use when Python code uses conditional imports, TYPE_CHECKING import branches, local imports that hide dependency problems, Ruff local-import diagnostics, or cyclic import diagnostics that need an ETHOS-grounded repair.