"SpecForge is a spec-driven roadmap and agent governance platform that eliminates contract drift, hallucinated implementation, and uncontrolled mutation in AI-assisted development. "
SpecForge is a structured AI governance and specification platform designed to bring order to the chaotic intersection of strategic vision and AI-assisted execution. By enforcing rigid contracts, schema validation, and full-lifecycle traceability, SpecForge ensures that LLM-generated artifacts align perfectly with business requirements and technical constraints. It bridges the gap between high-level strategy and verifiable implementation, providing a governed environment for responsible AI integration.
In the era of AI-driven development, engineering teams face significant challenges:
- Artifact Inconsistency: LLMs frequently generate artifacts that vary in style, structure, and quality, leading to technical debt.
- Specification Drift: As implementations evolve, they often drift away from original specifications without detection.
- Unenforceable Roadmaps: High-level roadmaps often lack the technical "teeth" to enforce contracts during execution.
- Reactive Governance: Compliance and validation are often treated as afterthoughts rather than being embedded in the development workflow.
SpecForge solves these problems by providing:
- Structured Spec Modeling: Moving from loose natural language to formal, schema-backed specifications.
- Roadmap-Driven Contracts: Automatically generating enforceable technical contracts directly from strategic roadmap units.
- Governed LLM Augmentation: Allowing LLMs to assist in creation while keeping them under strict schema and policy constraints.
- Full Traceability: Maintaining an immutable audit log and lineage for every variable, contract, and spec.
A Project is the primary lifecycle container in SpecForge. It encapsulates all roadmap items, technical contracts, validation rules, and artifacts for a specific business initiative. Projects define the governance boundary and tech stack context.
Strategic execution units that define the "What" and the "How". Each roadmap item includes:
- Goals & Business Context: Clear definitions of success.
- Deliverables: Specific technical artifacts required.
- Contracts: Technical interface definitions.
- Validation Logic: Rules that must pass before the item is considered complete.
- LLM Bootstrap: Users may optionally invoke an LLM to auto-populate roadmap fields, validation schemas, and contracts. These proposals are subjected to SpecForge’s governance engine and human review before acceptance.
Formal, enforceable definitions of system interfaces including REST APIs, GraphQL schemas, CLI commands, and internal function signatures. Contracts define:
- Input/Output Schemas: Rigorous JSON schema enforcement.
- Constraints: Range, regex, and relational constraints.
- Error Handling: Standardized failure modes and messages.
The heart of SpecForge governance. It performs:
- JSON Schema Enforcement: Ensuring all artifacts match their defined models.
- Intelligence Alignment: Detecting conflicts between related contracts or roadmap items.
- Dependency Graph Validation: Checking that all referenced variables and contracts resolve correctly.
- Cross-Artifact Consistency: Detecting drift between different versions or related specifications.
SpecForge moves beyond human-readable docs toward deterministic build artifacts. Roadmap items can be exported as self-contained packages that include all context, contracts, and validation rules required for an autonomous coding agent to implement the feature with zero ambiguity.
SpecForge treats LLMs as powerful assistants, not authoritative sources. All LLM-generated output is schema-checked before persistence. Furthermore, the AI Self-Evaluation engine scores the quality, ambiguity, and security of AI-proposed specs before they are presented for human review.
| Feature | Description |
|---|---|
| Structured Modeling | Define features and requirements using domain-driven schemas. |
| Project Bootstrap | Automated ingestion of existing codebases to generate structured intelligence. |
| Build Artifact Export | Generate deterministic, agent-ready implementation packages (.specforge-build). |
| Intelligence Alignment | Detect logical misalignments and schema conflicts across the entire project. |
| Schema Enforcement | Real-time validation of all technical contracts and variables. |
| Contract Builder | Visual and code-based editors for REST, Event, and CLI contracts. |
| Drift Tracking | Automated detection of breaking changes and spec-to-code gap. |
| Audit Logs | Immutable history of all governance decisions and artifact mutations. |
| UI Roadmap Engine | Visual, deterministic UI specification with state machines and component trees. |
| Intelligence Scoring | Real-time metrics on spec completeness, readiness, and AI trust. |
| MCP Server Integration | Deep integration with Model Context Protocol for Agentic IDEs (Cursor, Windsurf) and Gen-AI assistants. |
SpecForge allows any Roadmap Item to generate a fully self-contained, agent-ready implementation package. These artifacts eliminate the ambiguity between a specification and its execution.
- Deterministic Bundles: Exports include all contextual info, contracts, variables, and governance constraints in a structured
.specforge-buildformat. - AI Implementation Prompts: Automatically generates hyper-specific prompts for coding agents (Cursor, Anti-Gravity), enforcing zero-assumption development.
- Refinement Loops: Includes verification instructions and test generation requirements to guide agents through iterative self-correction.
The Bootstrap Layer allows teams to quickly onboard existing projects by ingesting their current state and generating a formal SpecForge foundation.
- Codebase Ingestion: Automatically scans repositories to map modules, APIs, and data models.
- Structured Intelligence Snapshots: Generates architecture scores, contract density metrics, and risk assessments based on existing code.
- Vision-to-Verifiable Execution: Bridges the gap for legacy projects, bringing them under the strict governance of SpecForge contracts.
Modern systems are complex webs of dependencies. SpecForge ensures that no logic change in one module silently breaks another.
- Collision Detection: Detects when two contracts define conflicting schemas for the same entity.
- Alignment Scoring: A real-time metric indicating how well specifications across the project harmonize with each other.
- Self-Evaluating AI: All AI suggestions are scored for ambiguity and security before human review, ensuring the highest "Trust Score" for new specifications.
SpecForge introduces the official UI-FRE, a deterministic specification engine for user interfaces. It eliminates the "handover gap" by modeling UI features as formal technical contracts.
- Deterministic Component Trees: Model UI hierarchies as strictly-typed component nodes.
- State Machine Diagramming: Define interactive flows (idle, loading, error, success) with visual and interaction mutations.
- Implementation Scoring: Real-time "Readiness Score" based on contract compliance, accessibility audits (A11y), and backend binding accuracy.
- Multi-Artifact Export: Generate implementation stubs for React/TS, Figma AI "Make" prompts, and Storybook stories directly from the spec.
SpecForge is built with a modern, high-performance stack designed for reliability and low-latency validation.
graph TD
subgraph Client
UI["Web Frontend (React + Tailwind)"]
IDE["Agentic IDE (Cursor, Windsurf)"]
end
subgraph "SpecForge MCP"
MCP["MCP Server"]
end
subgraph "SpecForge Core"
API["REST API (Go Backend)"]
subgraph Engine ["Intelligence & Governance Engine"]
VAL["Validation Engine"]
DRIFT["Drift Engine"]
LLM["LLM Layer (OpenAI/Claude)"]
AUDIT["Audit Log Service"]
BOOT["Bootstrap Layer"]
end
end
subgraph Persistence
DB[("PostgreSQL + S3 (Artifact Registry)")]
end
IDE --> MCP
MCP --> API
UI --> API
API --> VAL
API --> DRIFT
API --> LLM
API --> AUDIT
API --> BOOT
VAL & DRIFT & LLM & AUDIT & BOOT --> DB
- Frontend: React-based SPA with high-interactive visualization panels for dependency graphs and intelligence dashboards.
- Backend API: Go-based services with a high-performance Intelligence Engine for real-time validation, alignment, and scoring.
- Bootstrap Layer: Specialized ingestion engine for codebase analysis and snapshot generation.
- Validation & Alignment: AUTHORITATIVE logic engine for JSON schema enforcement and cross-artifact consistency.
- Artifact Registry: Versioned, immutable storage of all specifications, build artifacts, and snapshots.
- Initialize: User creates a new Roadmap Item (e.g., "Add User Authentication").
- Trigger AI: User enables "Populate with LLM".
- Context Assembly: SpecForge gathers project context, existing variable registries, and tech stack constraints.
- Generation: LLM returns a structured JSON proposal containing description, requirements, and contract stubs.
- Validation: The system runs the proposal through the Validation Engine (Schema check, Dependency check).
- Human-in-the-loop: User reviews the diff, edits where necessary, and approves.
- Persistence: The validated artifact is committed to the registry with full lineage.
{
"roadmap_item": {
"title": "API Gateway Rate Limiting",
"type": "FEATURE",
"priority": "HIGH",
"validation_rules": [
{ "name": "MaxRequestLimit", "type": "RANGE", "config": { "max": 1000 } }
]
},
"proposed_contracts": [
{
"type": "REST",
"path": "/api/v1/limits",
"method": "GET",
"output_schema": { "type": "object", "properties": { "remaining": { "type": "integer" } } }
}
]
}Navigate to the Workspaces dashboard and select "New Project". Define your tech stack and repository settings. Governance policies (strictness levels) are defined at this stage.
Inside the Project View, click "Add Roadmap Item". You can choose to manually enter details or use the AI Bootstrap feature to generate a complete draft from a simple prompt.
Use the Contract Editor to define your technical interfaces. Every field is typed and backed by the Variable Registry, ensuring consistency across the entire project.
Define the "Rules of Engagement" in the Validation Rules Panel. Rules can range from simple schema checks to complex cross-field dependencies.
Visualize how your contracts and items relate. The graph highlights missing dependencies or mismatched schemas in real-time.
- Dashboard: High-level health and "Intelligence Score" across all features.
- Roadmap Panel: Timeline view of execution units.
- Audit Logs: View exactly who changed what, and why.
SpecForge is engineered for high-compliance and mission-critical environments.
- Regulated Environments: Direct mapping of specifications to regulatory controls (SOC2, HIPAA, GDPR).
- AI Governance: Transparent and auditable use of LLMs in the SDLC.
- Security Specification: Modeling security controls as first-class, verifiable contracts.
- ColdWater Security Integration: SpecForge is the preferred platform for enterprise security teams, such as those at ColdWater Security, to model and enforce defensive architectures.
- Go 1.25+
- Node.js 24+
- PostgreSQL 16+
- Docker & Docker Compose (for containerized setup)
- OpenAI / Anthropic API Key (for Intelligence features)
You can run the entire SpecForge stack (Frontend, Backend, and Database) using Docker Compose.
cp backend/.env.example backend/.env
# Update .env with your OpenAI/Anthropic API keys if needed
docker-compose up --buildThis will automatically:
- Start a PostgreSQL 16 database.
- Run all database migrations.
- Start the Go Backend on
http://localhost:8080. - Start the React Frontend on
http://localhost:5173.
cd backend
cp .env.example .env
# Ensure PostgreSQL is running and update DATABASE_URL in .env if necessary
go mod download
go run cmd/api/main.gocd frontend
npm install
npm run devWe welcome contributions that align with our Validation-First philosophy.
- Schema First: Any new feature must start with a domain model update.
- Tests Required: All logic changes, especially in the Validation or Drift engines, require 100% test coverage.
- Documentation: README and OpenAPI specs must be updated alongside code.
- Hallucination Mitigation: All LLM outputs are treated as untrusted until validated by the schema engine and human review.
- Immutable Audit Logs: All state changes are recorded in an append-only log.
- Contract Immutability: Once a contract is used in production (approved), it must follow strict versioning and deprecation policies.
- [COMPLETED] Build Artifact Export Engine: Agent-native deterministic exports.
- [COMPLETED] Project Bootstrap Intelligence: Codebase ingestion and auto-spec generation.
- [COMPLETED] Intelligence Alignment Engine: Cross-feature conflict detection.
- [COMPLETED] UI Feature Roadmap Engine: Deterministic UI spec modeling and scoring.
- [COMPLETED] MCP Server Integration: Full support for Model Context Protocol allowing Gen-AI IDE integration.
- Figma Live Sync: Bi-directional synchronization between design and technical artifacts.
- Interactive State Simulation: Visual playback of state machine transitions in the dashboard.
- Spec Diff Visualizer: Advanced visual comparison for complex JSON schemas.
- Automated Compliance Mapping: Automatically tagging specs with compliance controls (SOC2, etc.).
- Policy-as-Code Integration: Enforcing organizational rules via OPA/Rego.
© 2026 SpecForge Platform. All Rights Reserved.