Framework-agnostic frontend engineering rules for human developers and AI coding agents.
The repository defines user-facing behavior, interaction decisions, and validation expectations. It contains documentation and evaluation material rather than UI components or framework-specific implementation code.
Read in this order:
AGENTS.mdfor the execution workflow and precedence rules;CORE_RULES.mdfor the cross-task baseline;- the task map in
docs/README.md; - the applicable rules in
FRONTEND_SPEC.mdand relevant detailed chapters; docs/10-validation-checklist.mdbefore delivery.
The full specification contains stable IDs such as FG-OVERLAY-002. Cite applicable IDs in plans, reviews, tests, and delivery reports, together with observable evidence.
Pin a released version and place it where the agent can read it locally, for example:
docs/frontend-guidelines/
A submodule, subtree, or vendored copy is preferable to a remote-only link. Then:
- merge
templates/PROJECT_AGENTS_SNIPPET.mdinto the project-levelAGENTS.md; - complete
templates/PROJECT_FRONTEND_PROFILE.md; - use the task-plan and change-report templates for material changes.
See examples/adoption.md for a compact end-to-end example.
AGENTS.md: AI execution contractCORE_RULES.md: minimum cross-task baselineFRONTEND_SPEC.md: complete normative rule catalogdocs/: rationale, failure modes, and acceptance guidancetemplates/: project profile, task plan, and delivery reportexamples/: framework-neutral examples and adoption flowevals/: cases for evaluating agent decisions
Read CONTRIBUTING.md. Rule proposals should describe a real user problem, the specification gap, observable acceptance criteria, and compatibility impact.
Run repository validation before submitting:
python3 scripts/validate.py