frontend-patterns/SCOPE.md

SCOPE

Reviewed: 2026-09-28 · Scope: ongoing research and a versioned knowledge base.

Purpose and deliverable

Develop a framework-independent frontend architecture pattern language for human and agent reference, following INTENT.md. Its intended practical outcome is an explainable path from a frontend problem and its forces to compatible patterns and implementation options.

The deliverable is a versioned Markdown knowledge base with structured metadata, traceable evidence and transparent revision history. The language evolves through bounded investigations and review gates; it has no final completion date. Completing an experiment or milestone does not establish a mature language.

In scope

• canon/: responsibilities, forces, scales, relations and the metadata contract, including testing whether the proposed vocabulary explains contrasting cases.
• research/: identifiable sources, claim-level interpretations, candidate patterns, counterexamples, pinned implementation studies and bounded executable experiments.
• patterns/: evidence-qualified canonical records when admission is justified; research candidates remain separately identified until then.
• sequences/ and architectures/: decision ordering, compatible configurations, alternatives and resulting contexts, exercised through contrasting walkthroughs.
• technologies/: descriptive mappings from patterns to implementation approaches, including their constraints and limits.
• templates/, scripts/ and the infospace-bench adapter: authoring, metadata checks and reproducible research evaluation. Live model execution uses llm-connect.
• research/evolution/ and history/: evidence profiles, immutable snapshots, prospective questions and effort, supersession, learning outcomes and dated assessments. Track applicability, trustworthiness, usefulness, coherence, maintainability and learning efficiency separately.

These areas include intended capabilities; directory presence is not evidence that the corresponding capability is implemented or validated.

Current capabilities and limits

As of 2026-09-28, eighteen developed research candidates remain draft hypotheses; none is canonically admitted. Sources and claim-level reviews, metadata contract checks, an infospace-bench adapter and bounded runtime research fixtures exist. Four walkthroughs are synthetic drafts or provisional studies, not validated architecture recipes. Six further validation cases remain planned.

The frozen development pilot scored 18/24 in each presentation condition after submitted model-review adjudication. It establishes neither a naming benefit nor equivalence; independent human validation and decision-time measurement remain open.

Evolution profiles, a baseline, learning ledger, snapshots and a generated report are implemented. Eight prospective cycles (seven effort-accounted) belong to seven different comparison groups: there is no comparable window for a diminishing-returns claim. Evidence types, implementation families and review provenance remain distinct. Static source mechanisms and authored runtime fixtures do not establish complete pattern recurrence, production fitness or independent human confirmation.

Canonical admission, M1/M2 acceptance, demonstrated generativity, full graph export, held-out retrieval evaluation remain open. Scoped mechanism mappings are available in the 0.1 preview; full implementation contracts remain unverified. There is research runtime code, but no maintained frontend application or framework is offered as the product. See the evolution report for current counts and evidence limits; REVIEW.md preserves the initial 2026-09-23 review and is not a current capability inventory.

Research boundaries and exclusions

• Do not build or operate a general frontend application, framework, router or release platform as the deliverable. Fixtures exist to investigate claims.
• Do not rank frameworks, prescribe a universal architecture or turn the language into a component cookbook or collection of tool-specific idioms.
• Do not treat candidate counts, source counts, passing assertions or model agreement as maturity, independent recurrence or proof of usefulness.
• Do not infer accessibility conformance, production guarantees or upstream defects from bounded DOM experiments or partial source inspection.
• Do not combine viability dimensions into a maturity percentage, infer historical effort from commits, or pool incomparable studies to claim learning saturation.

Work ownership and review boundaries

WP-0002 delivered seed evidence and the connected corpus (finished; M1 revise/defer). WP-0009 owns independent M1 validation. WP-0003 owns contrasting streams, broader walkthroughs, ontology decisions and M2. WP-0004 is blocked follow-up for graph, held-out retrieval and real-repository applicability, subject to its recorded gates. WP-0006 owns ongoing learning visibility, review dates and comparable-cycle measurement. Actionable work stays in workplans; this file describes boundaries, not a second backlog.

WP-0007 packages a finite 0.1 research preview: curated candidate reference paths, two core sequences, scoped mappings and a candidate export. This is a bounded research preview, not a replacement for M1/M2/M3.

The 2026-09-27 scope–intent assessment records the current gaps and recommended priorities. Future assessments should preserve that historical judgment and cite changed evidence.

Ecosystem reuse and operational learning

The user-directed WP-0008 connects qualified reference use to evidence-bearing consumer receipts and bounded autonomous improvement. It owns discovery, instrumentation and execution integration; monthly human review owns operating-policy decisions. The capability entry is prepared; automatic collection and three activity-core reporting schedules passed scheduled smoke checks; the first natural daily run failed after a worker image change removed the domain resolver. The approved grant-preserving repair is deployed and all three scheduled sink smokes pass; natural recurrence and October 1 evidence remain due. Hosted consumer integration and admitted pattern-editing execution remain blocked; there are no enrolled consumers yet. This operational loop can improve draft guidance without claiming M1/M2/M3 or independent-human validation. See the operating design.

Registration

• Repository: frontend-patterns
• Category: research
• Domain / topic: infotech / canon
• Workplan prefix: FEP-WP

The topic groups this research with semantic coordination work; it does not imply that the proposed language is an adopted fleet standard.