Intent Integrity Kit Analyze

Non-destructive cross-artifact consistency analysis across spec.md, plan.md, and tasks.md.

Operating Constraints

• READ-ONLY (exceptions: writes analysis.md and .specify/score-history.json). Never modify spec, plan, or task files.
• Constitution is non-negotiable: conflicts are automatically CRITICAL.

User Input

$ARGUMENTS

You MUST consider the user input before proceeding (if not empty).

Constitution Loading

Load constitution per constitution-loading.md (basic mode — ERROR if missing). Extract principle names and normative statements.

Prerequisites Check

1. Run: bash .tessl/tiles/tessl-labs/intent-integrity-kit/skills/iikit-07-analyze/scripts/bash/check-prerequisites.sh --phase 07 --json Windows: pwsh .tessl/tiles/tessl-labs/intent-integrity-kit/skills/iikit-07-analyze/scripts/powershell/check-prerequisites.ps1 -Phase 07 -Json

2. Derive paths: SPEC, PLAN, TASKS from FEATURE_DIR. ERROR if any missing.

3. If JSON contains needs_selection: true: present the features array as a numbered table (name and stage columns). Follow the options presentation pattern in conversation-guide.md. After user selects, run:

   bash .tessl/tiles/tessl-labs/intent-integrity-kit/skills/iikit-07-analyze/scripts/bash/set-active-feature.sh --json <selection>

   Windows: pwsh .tessl/tiles/tessl-labs/intent-integrity-kit/skills/iikit-07-analyze/scripts/powershell/set-active-feature.ps1 -Json <selection>

   Then re-run the prerequisites check from step 1.

4. Checklist gate per checklist-gate.md.

Execution Steps

1. Load Artifacts (Progressive)

From spec.md: overview, requirements, user stories, edge cases. From plan.md: architecture, data model refs, phases, constraints. From tasks.md: task IDs, descriptions, phases, [P] markers, file paths.

2. Build Semantic Models

• Requirements inventory (functional + non-functional)
• User story/action inventory with acceptance criteria
• Task coverage mapping (task -> requirements/stories)
• Plan coverage mapping (requirement ID → plan.md sections where referenced)
• Constitution rule set

3. Detection Passes (limit 50 findings)

A. Duplication: near-duplicate requirements -> consolidate B. Ambiguity: vague terms (fast, scalable, secure) without measurable criteria; unresolved placeholders C. Underspecification: requirements missing objects/outcomes; stories without acceptance criteria; tasks referencing undefined components D. Constitution Alignment: conflicts with MUST principles; missing mandated sections. For each principle, report status using these exact values:

• ALIGNED — principle satisfied across all artifacts
• VIOLATION — principle violated (auto-CRITICAL severity) E. Phase Separation Violations: per phase-separation-rules.md — tech in constitution, implementation in spec, governance in plan F. Coverage Gaps: requirements with zero tasks; tasks with no mapped requirement; non-functional requirements not in tasks; requirements not referenced in plan.md

Plan coverage detection: Scan plan.md for each requirement ID (FR-xxx, SC-xxx). A requirement is "covered by plan" if its ID appears anywhere in plan.md. Collect contextual refs (KDD-x, section headers) where found.

G. Inconsistency: terminology drift; entities in plan but not spec; conflicting requirements

G2. Prose Range Detection: Scan tasks.md for patterns like "TS-XXX through TS-XXX" or "TS-XXX to TS-XXX". Flag as MEDIUM finding: "Prose range detected — intermediate IDs not traceable. Use explicit comma-separated list."

H. Feature File Traceability (when FEATURE_DIR/tests/features/ exists): Parse all .feature files in tests/features/ and extract Gherkin tags:

• @FR-XXX — functional requirement references
• @SC-XXX — success criteria references
• @US-XXX — user story references
• @TS-XXX — test specification IDs

H1. Untested requirements: For each FR-XXX and SC-XXX in spec.md, check if at least one .feature file has a corresponding @FR-XXX or @SC-XXX tag. Flag any FR-XXX or SC-XXX without a matching tag as "untested requirement" (severity: HIGH).

H2. Orphaned tags: For each @FR-XXX or @SC-XXX tag found in .feature files, verify the referenced ID exists in spec.md. Flag tags referencing non-existent IDs as "orphaned traceability tag" (severity: MEDIUM).

H3. Step definition coverage (optional): If tests/step_definitions/ exists alongside tests/features/, run verify-steps.sh to check for undefined steps:

bash .tessl/tiles/tessl-labs/intent-integrity-kit/skills/iikit-07-analyze/scripts/bash/verify-steps.sh --json "FEATURE_DIR/tests/features" "FEATURE_DIR/plan.md"

If status is BLOCKED, report undefined steps as findings (severity: HIGH). If DEGRADED, note in report but do not flag as finding.

4. Severity

• CRITICAL: constitution MUST violations, phase separation, missing core artifact, zero-coverage blocking requirement
• HIGH: duplicates, conflicting requirements, ambiguous security/performance, untestable criteria
• MEDIUM: terminology drift, missing non-functional coverage, underspecified edge cases
• LOW: style/wording, minor redundancy

5. Analysis Report

Output to console AND write to FEATURE_DIR/analysis.md:

## Specification Analysis Report

| ID | Category | Severity | Location(s) | Summary | Recommendation |
|----|----------|----------|-------------|---------|----------------|

**Constitution Alignment**: principle name -> status (ALIGNED | VIOLATION) -> notes
**Coverage Summary**: requirement key -> has task? -> task IDs -> has plan? -> plan refs
**Phase Separation Violations**: artifact, line, violation, severity
**Metrics**: total requirements, total tasks, coverage %, ambiguity count, critical issues

**Health Score**: <score>/100 (<trend>)

## Score History

| Run | Score | Coverage | Critical | High | Medium | Low | Total |
|-----|-------|----------|----------|------|--------|-----|-------|
| <timestamp> | <score> | <coverage>% | <critical> | <high> | <medium> | <low> | <total_findings> |

5b. Score History

After computing Metrics in step 5, persist the health score:

1. Compute health score: score = 100 - (critical*20 + high*5 + medium*2 + low*0.5), floored at 0, rounded to nearest integer.
2. Read .specify/score-history.json. If the file does not exist, initialize with {}.
3. Append a new entry for the current feature (keyed by feature directory name, e.g. 001-user-auth):

   { "timestamp": "<ISO-8601 UTC>", "score": <n>, "coverage_pct": <n>, "critical": <n>, "high": <n>, "medium": <n>, "low": <n>, "total_findings": <n> }

4. Write the updated object back to .specify/score-history.json.
5. Determine trend by comparing the new score to the previous entry (if any):
   • Score increased → ↑ improving
   • Score decreased → ↓ declining
   • Score unchanged or no previous entry → → stable
6. Display in console output: Health Score: <score>/100 (<trend>)
7. Include the full score_history array for the current feature in analysis.md under the Health Score line and Score History table added in step 5.

6. Next Actions

• CRITICAL issues: recommend resolving before /iikit-08-implement
• LOW/MEDIUM only: may proceed with improvement suggestions

7. Offer Remediation

Ask: "Suggest concrete remediation edits for the top N issues?" Do NOT apply automatically.

Operating Principles

• Minimal high-signal tokens, progressive disclosure, limit to 50 findings
• Never modify files, never hallucinate missing sections
• Prioritize constitution violations, use specific examples over exhaustive rules
• Report zero issues gracefully with coverage statistics

Dashboard Refresh

Regenerate the dashboard so the pipeline reflects the analysis results:

bash .tessl/tiles/tessl-labs/intent-integrity-kit/skills/iikit-07-analyze/scripts/bash/generate-dashboard-safe.sh

Windows: pwsh .tessl/tiles/tessl-labs/intent-integrity-kit/skills/iikit-07-analyze/scripts/powershell/generate-dashboard-safe.ps1

Next Steps

You MUST read model-recommendations.md, check the expiration date (refresh via web search if expired), detect the agent via env vars, and include a model switch tip in the output below if the next phase needs a different model tier.

Analysis complete!
- CRITICAL issues found: resolve, then re-run /iikit-07-analyze
- No CRITICAL: /clear then /iikit-08-implement (implementation benefits from max context)
Tip: <model switch suggestion if tier mismatch, omit if already on the right model>
- Dashboard: file://$(pwd)/.specify/dashboard.html (resolve the path)