TOPOLOGY.adoc

Empty-linter topology and readiness

Implemented product paths

stdlib/ByteDetector.affine ──(AffineScript, build-only)──► src/core/ByteDetector.bun.js
                                                              (legacy minimum core)
src/core/UnicodeData.bun.js ──► src/core/ScalarScanner.bun.js ──┐
src/core/Render.bun.js                                          │
src/core/ScannerIO.bun.js  ── enumeration, fatal UTF-8, routing │
src/containers/ (zip · ooxml · pdf · hidden-style)              ├─► scanFile()
src/core/Settings.bun.js  ── config.ncl (Ni subset) / .json     │
                                                                ▼
                              ┌────────────── scripts/empty-lint-ci.js (stable CI gate)
                              ├────────────── src/cli/Main.bun.js (audit · show · plan ·
                              │              approve · apply · verify · tui)
                              └────────────── src/tui/ (interactive UI)
                                                                │
              Repair state machine (src/core/Repair.bun.js) ◄───┘
              audit → propose → approve → apply-to-copy → verify → refuse
                                                                │
              schemas/ (diagnostic · repair-plan · provenance · rescan)
              scripts/sync-downstream.js ─► userscript + VS Code extension

Every record the product emits is versioned JSON validated in-repo against the shipped schemas. The dogfood gate plants positive and negative controls before trusting the repository result, never suppresses scanner errors, and never modifies files.

Honest readiness

Capability	Status	Evidence or remaining work
Basic code-point detector (legacy minimum)	Implemented	AffineScript type-check plus its Bun test suite
Expanded detector catalogue (bidi, tags, VS, fillers, controls, zalgo)	Implemented, tested	tests/UnicodeData_test.js, tests/ScalarScanner_test.js; scalar columns, byte offsets, UTF-8 bytes, escaped context
Settings (config.ncl active + validated)	Implemented, tested	tests/Settings_test.js, tests/Cli_test.js — distinct failures for invalid values and unsupported Nickel features
Safe repair pipeline (plan/approve/apply-to-copy/verify/refuse + patch provenance + rescan)	Implemented, tested	tests/Repair_test.js, tests/IetfIncident_test.js
Machine-readable schemas	Implemented, tested	schemas/; real emitted records validated in tests/SchemaValidator_test.js
TUI (scan/filter/inspect/plan/approve/apply-to-copy)	Implemented; IO shell thin	Pure model + renderer + key decoder tested in tests/Tui_test.js
Container detectors (OOXML/PDF/spreadsheet/hidden styles)	Implemented, tested	tests/Containers_test.js with in-memory synthetic documents
Repository audit gate (stable entry point, exit codes)	Implemented	Seven+ end-to-end controls in tests/empty_lint_ci_test.js plus expansion
Downstream drift control (userscript, VS Code extension)	Implemented, gated	scripts/sync-downstream.js --check in CI
IETF incident acceptance fixture	Implemented, tested	tests/IetfIncident_test.js (every issue-#74 bullet asserted)
Expanded catalogue in AffineScript	Not implemented	Upstream compiler work keeps the reviewed JS engine as the interim source
Idris-backed scanner/repair proof	Not demonstrated	Proof obligations must be connected to the active implementation
Computed-style / JS-driven hidden-text analysis	Not implemented	Markup detector is a documented pattern detector only
Suite integration (Spline, Cleave, ForthWall)	Proposed / proposal-only	Groove manifest records the honest statuses; no duplicate engine created

No overall completion percentage is published: combining implemented and aspirational components into one percentage would imply evidence that does not exist.