Constitutional Governance for Technical Decisions
Immutable, structured, and enforceable architecture decision system. 40+ AI tools, 10 industry templates. AI assists, humans decide.
The problem we solve
Decisions get lost
Scattered across Slack, meeting notes, or someone's memory. New engineers don't know why the system was built a certain way.
ADRs don't work
Inconsistent format, no validation, not searchable by AI, and no enforcement.
AI lacks context
AI assistants start from zero every session. They don't know your team's rules and may suggest contradictory things.
MANTRA becomes a constitutional memory for your AI assistant — immutable decisions, structured, and automatically enforced.
Measurable outcomes
Not just features, but real impact for your team.
Everything you need
From decision creation to automatic enforcement, all in one system.
4×4 Decision Matrix
4 Domains (INT, ARCH, CTL, EVO) × 4 Aspects = 16 structured and consistent decision categories.
Immutable Records
Decisions cannot be modified or deleted. Evolution only through supersedes and semantic versioning.
3-Gate Validation
Gate 1: Schema. Gate 2: AI quality check. Gate 3: Human approval. No shortcuts.
Smart Retrieval & MICS
Context-aware search with 7-stage AI pipeline. Trigger-based suggestions and knowledge graph.
Constraint Enforcement
Regex checks, import rules, violation tracking, and CI/CD integration for automated enforcement.
40+ MCP AI Tools
Integrated with Claude Code, Cursor, Windsurf through Model Context Protocol.
10 Industry Templates
Pre-built templates for Software, Healthcare, Government, Finance, and more.
Document & IDE Export
14 document types + 9 IDE rule formats (.cursorrules, CLAUDE.md, Copilot).
Why MANTRA?
Compare with other approaches to managing architecture decisions.
| No Governance | Traditional ADR | MANTRA | |
|---|---|---|---|
| Decisions stored | Lost in Slack/email | Markdown files in repo | Immutable database |
| AI accessible | ✗ | ✗ | ✓ 40+ MCP tools |
| Quality validation | ✗ | ✗ | ✓ 3-stage automatic |
| Rule enforcement | ✗ | ✗ Manual | ✓ Automatic |
| Semantic search | ✗ | ✗ | ✓ AI + keyword |
| Engineer onboarding | Weeks | Read docs yourself | 5-minute setup |
| Multi-industry | ✗ | ✗ | ✓ 10 templates |
| Document export | ✗ | Manual | ✓ 14 types + 9 IDE |
How it works
Three simple steps: draft, validate, store.
Draft
Create architecture decisions via MCP tool in Claude Code, or directly from the web dashboard.
Validate
3-gate pipeline: schema validation → AI quality check → human approval.
Store
Immutable record stored permanently. AI assistants automatically retrieve it when needed.
A powerful dashboard
Manage decisions, monitor compliance, and analyze from one place.
Works with AI assistants
40+ MCP tools to retrieve, review, and enforce decisions directly from your IDE.
// Automatically retrieve relevant decisions
mantra_retrieve({
query: "authentication JWT implementation",
file_path: "src/auth/middleware.ts"
})
// Automatically check code compliance
mantra_review({
file_path: "src/api/routes.py",
file_content: "..."
})Built by engineers, for engineers
MANTRA was designed by a team that understands the frustration of lost and unenforced architecture decisions.
Join the teams already managing their architecture decisions in a better way.
Start governing your decisions today
Free for 3 members and 25 decisions. Create your first architecture decision in minutes.