Skip to content

docs: x-FuSa spec v1.13.0 — evidence-artifact schemas + content-quality baseline - #82

Merged
SoundMatt merged 1 commit into
mainfrom
spec-artifact-schemas-quality-baseline
Jul 28, 2026
Merged

SoundMatt merged 1 commit into
mainfrom
spec-artifact-schemas-quality-baseline

Conversation

@SoundMatt

Copy link
Copy Markdown
Owner

Summary

Follows a cross-tool content audit (all six x-FuSa tools) run against real live output: traceability and unit-test coverage were uniformly strong, but several evidence artifacts were schema-shaped without being information-bearing — hundreds of near-identical FMEA rows, an untouched HARA template, an artifact with invalid JSON. A schema alone can't catch fabricated-looking content, so this PR does two things:

1. Formalizes real schemas for the previously tool-defined evidence artifacts, each grounded in the actual governing standard instead of an invented shape:

  • §1.2.5 .fusa-hara.json (MUST read/validate) — ISO 26262-3 Clause 6, using the operationalSituations[]/hazards[]/safetyGoals[] cross-referenced shape already independently converged on across the family and matching FuSaOps' own hara package.
  • §9.2 hara/fmea/tara/safety-case (mirrors how comp was formalized) — fmea per IEC 60812/AIAG-VDA Handbook with an actionPriority field superseding raw RPN; tara per ISO/SAE 21434 Clause 15 using SFOP impact categories; safety-case per the GSN Community Standard v3 with all six real node types.
  • §9.3 sas/sci — DO-178C §11.20/§11.16.

2. Adds §1.6, a cross-cutting content-quality baseline (MUST unless noted) applying to every free-text evidence artifact: no placeholder/template text, structural JSON validity required, no single hardcoded qualitative fallback applied regardless of the item's actual signature, real (non-keyword) file/function referents only, plus SHOULD-level provenance ("analysis": "heuristic"|"reviewed") and freshness recommendations.

Also adds a §1.5.2 SHOULD rule for requirement-text quality, grounded in ISO/IEC/IEEE 29148:2018 and EARS sentence patterns.

SpecVersion bumped 1.12.0 → 1.13.0.

Scope note: this is the spec-design step only — none of these six schemas is yet implemented by any tool against the new shape, and none is consumed by FuSaOps (that remains a future MINOR bump, tracked separately per-tool as needed).

Test plan

  • go build ./... / go vet ./... — clean
  • go test ./... — all packages pass
  • Verified the new HARA schema matches FuSaOps' existing hara/hara.go implementation (no code changes needed, doc-only change)

…ty baseline

Formalizes HARA/FMEA/TARA/safety-case/SAS/SCI from tool-defined
sketches into real field-level MUST/SHOULD schemas grounded in
ISO 26262-3, IEC 60812/AIAG-VDA, ISO/SAE 21434, the GSN Community
Standard, and DO-178C, following a cross-tool content audit that found
traceability/coverage strong but several artifacts schema-shaped
without being information-bearing (templated FMEA boilerplate, an
untouched HARA template, an artifact with invalid JSON).

Adds a new cross-cutting §1.6 content-quality baseline (bans
placeholder text, requires JSON validity, bans a single hardcoded
qualitative fallback, requires real referents) that a schema alone
can't enforce, and a §1.5.2 requirement-text quality rule grounded in
ISO/IEC/IEEE 29148 and EARS. Bumps SpecVersion 1.12.0 -> 1.13.0.

Signed-off-by: SoundMatt <SoundMatt@users.noreply.github.com>
@SoundMatt
SoundMatt merged commit 09aafee into main Jul 28, 2026
13 checks passed
@SoundMatt
SoundMatt deleted the spec-artifact-schemas-quality-baseline branch July 28, 2026 17:55
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant