Spike-suite-4 CLOSED 2026-05-25T15:50 BST

2 spikes; 1 outright VALIDATED (T1) + 1 VALIDATED-WITH-NOTE (T2); 2 COMMIT decisions; 0 DEFER; 0 REJECT.

T1 mkdocs-material 4-way + hosting → VALIDATED → COMMIT

  • mkdocs-material 9.7.6 PASSED all 8 success criteria empirically against 5 real TT specs (7046 lines)
  • 1.39s build / 2.74MB initial-load / TT frontmatter accepted unmodified / mermaid + LaTeX math render
  • Hugo + Docusaurus FAILED on TT frontmatter (date format / MDX braces respectively)
  • Astro Starlight viable backup but heavier (35s install / 793MB peak RAM)
  • Cloudflare Pages recommended hosting (TT core infra already locked)
  • T-file: /home/richardd/off-github/library/projects/inherit/T-spike-t1-mkdocs-material-2026-05-25.md (commit 8df0b8f)

T2 EARS notation 10×5 fit-test → VALIDATED-WITH-NOTE → COMMIT

  • Fit-test 8/10 Natural / 2 Forced (both Cedar) / 0 Awkward — PASSES ≥7 threshold → ADOPT
  • Hook v1 strict EARS opening: 100% FN against TT’s **FR-001**: System MUST X convention
  • Hook v2 (FR-NNN-prefix-strip + ‘System’ as 6th opening keyword): 13.5% FN / 6.2% FP — PASS ≤20%
  • 5 mandatory tunings locked for Stage 6 contract (see findings doc §2)
  • Scope limit: hook applies to .md spec/plan/tasks only; NOT Cedar/Catala/LinkML native files
  • T-file: /home/richardd/off-github/library/projects/inherit/T-spike-t2-ears-notation-fit-test-2026-05-25.md (commit aeb024f)

Findings doc + Path E audit bump

  • Findings: docs/superpowers/specs/2026-05-25-standardisation-spike-suite-4-findings.md v1.0
  • Path E audit v1.1 → v1.2 — Session B + Session D1 scope NOW UNBLOCKED for launch-prompt authoring
  • 4-suite combined: 9 COMMIT + 10 DEFER + 1 REJECT across 20 candidates; ~95% decision-locked
  • docs-strategy commit 1023889

Implications

  • Session B (Stage 10) launch-prompt can be authored with mkdocs-material 9.7.6 + Cloudflare Pages locked
  • Session D1 (Stage 6) launch-prompt can be authored with EARS adoption + hook v2 (TT-aware) locked
  • Out of scope for this session: actually authoring Session B / D1 launch-prompts (next session) + implementing the outcomes (Session B + D1 will execute)

Related: feedback_tasklist_replaces_active_work_log, project_tt_core_tool_stack_2026_05_22