Splice Frameworkv1.0.0
Read PLAYBOOK.md

Stop Managing Syntax. Start Orchestrating Context.

A context-driven engineering framework for the AI era. Compress development cycles from weeks to hours by replacing fragmented tasks with unified executable context.

Read the Playbook(.md)

The Paradigm Shift

The Old Way

  • Focuses on human typing speed as the primary output metric
  • Fragmented tickets scattered across sprints and boards
  • Measuring effort through story points and velocity
  • Treating syntax as the bottleneck to shipping

The Splice Way

  • Focuses on AI generation speed with human orchestration
  • Unified Context Blocks that carry complete intent
  • Measuring Context Ambiguity as the key metric
  • Treating context transfer as the only bottleneck

The Roles

Design

Context Architect

Owns business rules, constraints, and the Master Prompt. Never writes syntax. Transforms ambiguous requirements into precise, executable context that AI can generate from.

Execution

AI Orchestrator

The pilot in the IDE. Feeds context to the AI, stitches components together, and reprompts on syntax errors. Converts the Master Prompt into working code through iterative generation.

Review

Logic Auditor

Owns security, architecture scale, and edge cases. Reviews PRs for business logic flaws, completely ignoring syntax formatting. Guards the system against subtle defects.

Phase 0

The Research Phase

Highly ambiguous features start here. Before a single line of code is generated, the team reduces context ambiguity through structured research.

Input
User Interviews
Input
Competitor Analysis
Input
Technical Feasibility
Output
Master Research Document
Ambiguity Score: Low

This document lowers the ambiguity score before coding starts, ensuring the Context Architect has clear constraints to work with.

The Context Assembly Pipeline

Each domain expert injects their constraints asynchronously. The Context Architect merges them to resolve contradictions before AI generation begins.

UI/UX
frontend-context.md
API
backend-context.md
Testing
qa-context.md
DevOps
infra-context.md
Master Prompt
master-context.md
Output
Claude Code / Cursor IDE
Generating components...
Applying constraints...
Done.
Real-World Example

Create a Post Feature

A Linear-style workflow showing how a feature moves from context drafting to production.

Drafting Context
User Settings Page
Ambiguity: High
Ready for Vibe
Create a Post
Ambiguity: Low
Generating
Empty
Logic Audit
Image Upload
Ambiguity: Low
Live
Auth Flow
1

Architects write the DB schema and UI constraints in Markdown files. All edge cases documented.

2

The Linear task is scored as Low Ambiguity. Ready for AI generation.

3

The AI Orchestrator uses the Master Prompt in the IDE to generate the Server Actions and React components in 45 minutes.

4

The Logic Auditor reviews the PR to ensure the S3 upload is sanitized against malicious files.

5

Feature goes live. Total time: 2 hours instead of 2 weeks.

Executable DNA and Hyper-Learning

Zero-friction onboarding and the 20x Junior Acceleration effect.

Zero-Friction Onboarding

D1
Read the Rules

New engineer reads the .cursorrules and .clauderules files instead of dead wikis.

D2
Start Orchestrating

They sit as an AI Orchestrator, generating code from existing context under senior supervision.

The 20x Junior Acceleration

Instead of spending 3 days fighting basic syntax and typos to build a CRUD app, the AI generates the syntax instantly.

The Junior spends their time reading advanced, modular code and receiving architectural corrections from Seniors during the Logic Audit.

They learn software design, not just typing. Pattern recognition replaces rote memorization.

3 days
Traditional CRUD app
2 hours
With Splice