| name | Mark Kashef: Claude Claw — Agent SDK Bridge Architecture |
| description | Build a personal AI assistant that bridges any messaging platform to your existing Claude Code infrastructure via Anthropic's Agent SDK subprocess pattern — zero dual-entry, full capability access from your phone |
| version | 2.0 |
| format | completion-engine |
| workflows | 4 |
Mark Kashef: Claude Claw — Agent SDK Bridge Architecture
Mark Kashef, after trying and exhausting the OpenClaw/NanoClaw ecosystem (and building his own derivatives), discovered that the simplest and most powerful personal assistant pattern is bridging directly into your existing Claude Code instance using Anthropic's Agent SDK subprocess spawning. Instead of building a separate AI brain, you create a thin messaging bridge to the one that already works.
Available Workflows
| # | Workflow | Produces | Use When |
|---|
| bridge | Claude Claw Bridge & Memory Blueprint | A comprehensive technical architecture document including the 8-stage pipeline design and a 3-layer SQLite memory schema | Designing a production-ready bridge from a messaging platform to Claude Code with persistent, intelligent memory. |
| interactive | Self-Building Assistant Wizard | A self-executing 'Mega-Prompt' that interviews users to automatically configure and deploy their custom AI assistant | You need to package complex documentation and build instructions into a single, interactive setup artifact for end-users. |
| context-yield | Context Yield Optimization Strategy | A reusable context loading strategy that maximizes Claude Code output quality per token consumed for a specific workflow | You need to reduce context waste, fix late-session quality drift, or design prompt chains that compound understanding instead of re-loading files. |
| model-factory | Model-Card Onboarding & Prompt Factory | A 10-minute model migration guide with converted prompts, and/or a business-tailored per-modality production prompt library | A new model drops and your prompts need migrating from its model/system cards, or a business needs a tailored prompt suite generated from a research-built mega guide instead of a generic pack. |
Quick Reference
Execution Prompts (structure-pure v2)
3 deterministic practitioner prompts — each carries an Output Contract, Output Skeleton, and Quality Gate. When a deliverable matches one, Read it and honor its contract instead of improvising the output shape.
- Mark Kashef — Agent SDK Bridge Architect —
skills/mark-kashef-claude-claw/references/prompts-v2/bridge-architect.md
- Mark Kashef — 3-Layer Memory System Designer —
skills/mark-kashef-claude-claw/references/prompts-v2/memory-system-designer.md
- ======================================= —
skills/mark-kashef-claude-claw/references/prompts-v2/wizard-builder.md
