Analysis¶
These ten commands provide deep architectural analysis, from cross-ADR conflict detection through formal conformance checking to post-fix retrospectives. Each spawns a specialized agent with a single analytical focus.
/blueprint:impact: Cross-ADR Conflict Detection¶
Analyze an ADR's impact on other decisions and the codebase. Detects conflicts, dependencies, duplications, and affected components.
Syntax: /blueprint:impact N
Examples:
/blueprint:impact 12
# Check ADR-0012 for conflicts with all other accepted ADRs
/blueprint:impact 3
# Before accepting: does this conflict with anything?
Tip
Run impact before accepting any ADR that touches shared infrastructure (databases, auth, messaging). Cross-ADR conflicts are the most expensive bugs in architecture.
/blueprint:audit: Compliance Verification¶
Verify the codebase actually follows accepted decisions. The compliance auditor scans code for evidence of compliance and violations against each accepted ADR.
Syntax: /blueprint:audit [N]
Without an argument, audits all accepted ADRs. With a number, audits a specific one.
Verdicts: COMPLIANT, PARTIAL, VIOLATING, or INSUFFICIENT EVIDENCE.
Examples:
/blueprint:audit
# Full compliance audit of all accepted decisions
/blueprint:audit 7
# Check compliance for ADR-0007 specifically
Tip
Run audits after major feature branches merge. Individual PRs may each be compliant, but the aggregate can drift.
/blueprint:reflect: Reflexion Model Conformance¶
Compute a reflexion model (Murphy, Notkin, Sullivan, 1995) by comparing the intended architecture (from accepted ADRs + ARCHITECTURE.md) against actual source code structure. Reports convergences, divergences, and absences with file:line evidence.
Syntax: /blueprint:reflect
- Convergences: Code matches the intended architecture
- Divergences: Code contradicts the intended architecture
- Absences: Intended architecture elements not found in code
Examples:
Tip
reflect is the most rigorous conformance check Blueprint offers. Use it when you need formal evidence that the code matches the architecture, not just that it "seems right."
/blueprint:evidence: Epistemic Status Audit¶
Audit the epistemic status and temporal validity of evidence supporting accepted ADRs. Detects stale evidence, unverified AI-generated research, expired assumptions, and dead source URLs.
Syntax: /blueprint:evidence
Evidence levels:
- L0: Unverified, AI-generated or unsourced
- L1: Partially verified, some sources checked
- L2: Fully verified, all claims traced to authoritative sources
Examples:
/blueprint:evidence
# Full epistemic audit of all ADR evidence
# Output includes:
# - ADR-0003: Evidence STALE (benchmark from 2024, library has had 3 major versions since)
# - ADR-0007: Evidence L0 (AI-generated, no external sources cited)
# - ADR-0012: Evidence L2 VALID (all sources verified, within expiry)
/blueprint:map: Wardley Map Strategic Analysis¶
Generate a Wardley Map classifying components by evolution stage (Genesis, Custom, Product, Commodity). Detects strategic misalignment: building custom solutions for commodity problems, or using commodity tools for genesis-stage differentiators.
Syntax: /blueprint:map
Examples:
/blueprint:map
# Generate Wardley Map linked to ADRs
# Flags: "You're building a custom auth system (Genesis effort) for a problem
# that has commodity solutions (Auth0, Clerk). ADR-0005 should be revisited."
/blueprint:tradeoff: ATAM Quality Attribute Utility Trees¶
Generate ATAM-style quality attribute utility trees from accepted ADRs. Identifies sensitivity points, tradeoff points, risks, and non-risks.
Syntax: /blueprint:tradeoff
- Sensitivity point: A property where a small change in one attribute causes a large change in another
- Tradeoff point: A property that is a sensitivity point for multiple attributes
- Risk: A potentially problematic architectural decision
- Non-risk: A decision that appears sound under analysis
Examples:
/blueprint:tradeoff
# Generate utility tree with quality attributes, scenarios, and tradeoff identification
/blueprint:risk: Architecture Risk Heat Map¶
Generate a risk heat map by analyzing ADR coverage gaps, component complexity, git churn, coupling density, and missing governance.
Syntax: /blueprint:risk
Risk score formula: (Complexity x Churn x Coupling) / Governance
Components with high complexity, frequent changes, tight coupling, and no governing ADR are the highest risk.
Examples:
/blueprint:risk
# Heat map: "src/auth/ HIGH RISK: 12 files, 847 lines average, 23 cross-imports,
# high churn (42 commits in 30 days), NO governing ADR"
/blueprint:trace: ADR-to-Fitness Traceability¶
Show the traceability matrix between accepted ADRs and fitness functions. Reveals governance gaps: decisions that have no automated enforcement.
Syntax: /blueprint:trace
Examples:
/blueprint:trace
# Matrix output:
# ADR-0001 → fitness_001.sh [PASS]
# ADR-0003 → fitness_003.sh [PASS]
# ADR-0005 → [NO FITNESS FUNCTION] ← governance gap
# ADR-0007 → fitness_007.sh [FAIL]
/blueprint:retro: Post-Fix Retrospective¶
After any fix, evaluate whether it was a band-aid or systemic. Classifies root causes using config/taxonomy.toml, verifies proposed improvements against 3+ external sources, and produces a SYSTEMIC / BAND-AID / PARTIAL verdict.
Syntax: /blueprint:retro
Examples:
/blueprint:retro
# After fixing a bug:
# Verdict: BAND-AID
# Root cause: implicit contract between UserService and AuthMiddleware
# Proposed: /blueprint:new "explicit service contracts via TypeScript interfaces"
# External validation: 3/3 sources confirm (Microsoft, ThoughtWorks, Fowler)
Tip
The retro is Blueprint's institutional memory mechanism. When the same root cause class produces bugs repeatedly, the architecture has a gap that needs a decision, not another patch.
/blueprint:rearchitect: Supersede a Decision¶
Research a new approach, draft a superseding ADR, run impact analysis, and transition old ADR(s) to Superseded.
Syntax: /blueprint:rearchitect "topic"
Examples: