Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Creates, updates, validates, and displays the architectural DNA of a project through two shared documents: docs/specs/architecture.md (technology stack, architectural rules, security constraints, AI guardrails) and docs/specs/ontology.md (domain glossary / Ubiquitous Language). Use BEFORE brainstorm as a project setup step, or at any point in the SDD lifecycle to validate specs/tasks against architecture principles. Triggers on 'create constitution', 'update constitution', 'constitution check',
.claude/skills/giuseppe-trisciuoglio-constitution/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | 162% | 0% |
| case-04 | ✗→✓ | ▲ Improved | -55% | 0% |
| case-06 | ✗→✓ | ▲ Improved | 986% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 24% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 109% | 0% |
The Constitution skill manages the architectural DNA of a project through two shared documents:
| File | Purpose | |------|---------| | docs/specs/architecture.md | Technology stack, infrastructure, architectural rules, security constraints, AI guardrails | | docs/specs/ontology.md | Domain glossary (Ubiquitous Language) — terms, definitions, bounded contexts |
These files live at docs/specs/ and are shared across all specifications. Unlike a monolithic constitution.md, these are created/enriched by brainstorm (Phase 6.8.6) and spec-to-tasks (Phase 1.5).
$ARGUMENTS or user intent: create, update, check, or show.AskUserQuestion, then write the files using the templates below.Last Updated date.bash# Create constitution before first brainstorm /developer-kit-specs:constitution create # Validate a spec against architecture and ontology /developer-kit-specs:constitution check --target=docs/specs/001/2024-01-15--user-auth.md # Update the security constraints section /developer-kit-specs:constitution update --file=architecture --section=security # Show current constitution /developer-kit-specs:constitution show
| Scenario | Operation | |----------|-----------| | New project — define stack and domain language before first brainstorm | create | | Stack or security rules changed | update | | Validate a spec, task, or file against architecture and ontology | check | | Review current architecture and ontology | show |
Trigger phrases:
AskUserQuestion (domains, infrastructure, stack, data, style, rules)Template lookup order:
${CLAUDE_PLUGIN_ROOT}/templates/architecture.mdskills/constitution/references/architecture.md--file=architecture|ontology and --section=<name>Last Updated date--target=<path>)docs/specs/architecture.md and docs/specs/ontology.mdThe Constitution survives context rot through file-based storage:
docs/specs/architecture.md and docs/specs/ontology.mdFor detailed scenarios and recovery protocols, see references/context-rot-prevention.md.
Last Updated date on every changearchived/ — never let specs become stalebrainstorm (Phase 6.8.6) and spec-to-tasks (Phase 1.5)## Constitution Check Report
Target: <file path>
Date: YYYY-MM-DD
### Security Check (CWE/OWASP Compliance)
| Rule | Level | Status | Location | CWE/OWASP |
|------|-------|--------|----------|-----------|
| No SQL injection | CRITICAL | ✅ OK | - | CWE-89 |
### CWE Compliance Report
| CWE | OWASP | Status | Location |
|-----|-------|--------|----------|
| CWE-89 | A03 | ✅ OK | - |
### Architecture Check
| Rule | Status | Detail |
|------|--------|--------|
| Constructor injection | ✅ OK | - |
### Library Verification Check
| Library | Status | Detail |
|---------|--------|--------|
| bcrypt | ✅ OK | Using hash(password, 12) |
### Ontology Check
| Term | Status | Detail |
|------|--------|--------|
| "User" used consistently | ✅ OK | - |
### Summary
- CRITICAL violations: 0
- WARNING violations: 0
- Compliant rules: NFor detailed security patterns (CWE/OWASP mappings), see references/security-patterns.md.
[Session Start] → Read Constitution files
↓
[Optional] constitution create ← this skill (pre-brainstorm setup)
↓
brainstorm ← Constitution loaded before brainstorming
↓
spec-to-tasks ← Constitution validates spec
↓
task-implementation ← Constitution guardrails active
↓
task-review ← Constitution check validates
↓
[Session End] → Constitution files updated if neededRequired loading before:
specs.brainstorm — Validate requirements align with architecturespecs.spec-to-tasks — Check stack compatibilityspecs.task-implementation — Apply AI guardrailsspecs.task-review — Constitution check| File | Purpose | |------|---------| | references/architecture.md | Full architecture template | | references/ontology.md | Full ontology template | | references/security-patterns.md | CWE/OWASP patterns, verification format | | references/context-rot-prevention.md | Detailed scenarios and recovery protocols | | references/constitution-check-report.md | Complete report examples |
For complete templates and detailed reference material, consult the references/ directory.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 6,316 | 7,802 | +24% | 1 | 1 | 0% | 727 | 2,139 | +194% | 0 | 0 | — |
case-02 | fail→pass | 5,368 | 4,979 | -7% | 1 | 1 | 0% | 921 | 2,410 | +162% | 0 | 0 | — |
case-03 | fail→fail | 7,252 | 7,596 | +5% | 1 | 1 | 0% | 1,109 | 1,951 | +76% | 0 | 0 | — |
case-04 | fail→pass | 23,593 | 4,334 | -82% | 1 | 1 | 0% | 5,304 | 2,392 | -55% | 0 | 0 | — |
case-05 | fail→fail | 19,722 | 5,733 | -71% | 1 | 1 | 0% | 3,802 | 1,886 | -50% | 0 | 0 | — |
case-06 | fail→pass | 4,631 | 8,399 | +81% | 1 | 1 | 0% | 284 | 3,084 | +986% | 0 | 0 | — |
case-07 | fail→pass | 10,002 | 3,384 | -66% | 1 | 1 | 0% | 1,802 | 2,241 | +24% | 0 | 0 | — |
case-08 | fail→pass | 6,211 | 1,520 | -76% | 1 | 1 | 0% | 923 | 1,931 | +109% | 0 | 0 | — |
case-09 | fail→fail | 10,833 | 1,677 | -85% | 1 | 1 | 0% | 1,441 | 1,908 | +32% | 0 | 0 | — |
case-10 | pass→pass | 10,807 | 2,725 | -75% | 1 | 1 | 0% | 1,794 | 2,132 | +19% | 0 | 0 | — |
case-11 | fail→pass | 8,103 | 1,857 | -77% | 1 | 1 | 0% | 1,230 | 1,976 | +61% | 0 | 0 | — |
case-12 | pass→pass | 10,820 | 2,757 | -75% | 1 | 1 | 0% | 1,731 | 2,057 | +19% | 0 | 0 | — |
case-13 | fail→pass | 16,831 | 4,458 | -74% | 1 | 1 | 0% | 2,669 | 2,363 | -11% | 0 | 0 | — |
case-14 | fail→pass | 7,574 | 3,932 | -48% | 1 | 1 | 0% | 1,147 | 2,297 | +100% | 0 | 0 | — |
case-15 | fail→pass | 10,171 | 3,823 | -62% | 1 | 1 | 0% | 1,554 | 2,286 | +47% | 0 | 0 | — |
case-16 | fail→pass | 10,930 | 2,478 | -77% | 1 | 1 | 0% | 1,723 | 2,124 | +23% | 0 | 0 | — |
case-17 | fail→fail | 11,667 | 3,614 | -69% | 1 | 1 | 0% | 2,000 | 2,363 | +18% | 0 | 0 | — |
case-18 | pass→pass | 4,176 | 2,430 | -42% | 1 | 1 | 0% | 556 | 1,963 | +253% | 0 | 0 | — |
case-19 | pass→pass | 13,275 | 5,367 | -60% | 1 | 1 | 0% | 2,238 | 2,549 | +14% | 0 | 0 | — |
case-20 | pass→pass | 12,588 | 3,325 | -74% | 1 | 1 | 0% | 1,971 | 2,211 | +12% | 0 | 0 | — |
case-21 | fail→pass | 7,902 | 5,064 | -36% | 1 | 1 | 0% | 1,273 | 2,488 | +95% | 0 | 0 | — |
case-22 | fail→pass | 8,305 | 1,986 | -76% | 1 | 1 | 0% | 1,319 | 1,982 | +50% | 0 | 0 | — |
case-23 | fail→pass | 7,936 | 1,518 | -81% | 1 | 1 | 0% | 1,272 | 1,918 | +51% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 23 cases were attempted, and 20 counted toward the lift figure. The other 3 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +57 percentage points is the difference between those two pass rates over the 20 comparable cases. 1 case got worse with the skill loaded, and it is included in that figure.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.