Knowledge base¶
Reusable explanations of web sustainability methodology, evidence and tooling. Start here to understand what an audit result means; use the audit guide to run one. Implementation-specific commands and data contracts live in the workflow and its linked references.
Methodology and interpretation¶
- WSG and STAR: scope, publication status, source formats and evaluation limits.
- Scoped audit scoring: outcome mapping, impact weights, denominators and coverage.
- WSG auditability: current criterion classes, evidence needs and review routes; classification is distinct from implemented coverage.
- Digital Carbon Ratings: SWDM v4, thresholds and interpretation alongside WSG findings.
- Carbon.txt evidence: discovery, provider relationships, declarations and corroboration.
Measurement and tools¶
- Deterministic audit tools: existing collectors, browser instrumentation and capture protocols.
- Browser energy and traffic scenarios: counter boundaries, CrUX ranks and pageview estimation.
- Optional GreenFrame measurements: client-side scenario boundaries and remaining runner validation.
Agent engineering¶
-
Portable agent orchestration: the Wagtail CLI precedent, native subagents, durable handoffs and remaining design decisions.
-
Lessons from existing audit projects: patterns from claude-seo and accessibility-agents.
- Model providers and inference accounting: API contracts, versioning, confidence and measurement boundaries.
Project hypotheses belong in the feasibility study. Executed live results belong under reports; model experiments belong in the evaluation project. Source observations and limitations stay attached to each reference topic.