| name | Mark Kashef Agent Orchestration |
| description | Architect and orchestrate AI agent teams using Mark Kashef's specialized workflows for sequential handoffs, parallel synthesis, and hybrid grunt-to-architect models. |
| version | 2.0 |
| format | completion-engine |
| workflows | 4 |
Mark Kashef Agent Orchestration
Architect and orchestrate AI agent teams using Mark Kashef's specialized workflows. Transformed from MES 3.0 Deep Extraction, this skill enables you to treat LLMs as full corporate departments with embedded communication protocols and task blockades.
This skill implements the core genius of Mark Kashef: Structuring complex cognitive labor into discrete, intercommunicating agent roles. It focuses on the difference between simple, siloed sub-agents and complete communicating agent teams.
**Deploy
Available Workflows
| # | Workflow | Produces | Use When |
|---|
| multi | Multi-Agent Content & Strategy Engine | Integrated Project Dossier (e.g., Pitch Deck, Strategy Report, or Multi-Platform Content Suite) | You need to generate complex documents requiring either a step-by-step assembly line or a synthesized consensus from multiple expert perspectives. |
| architected | Architected Technical Build System | Verified Technical Implementation (e.g., Refactored Repository or Complex Feature Build) | You are performing large-scale code changes or system architecture where errors are expensive and require both deep repo-scale analysis and human oversight. |
| phase-gated | Phase-Gated Context Factory | A phase-gated execution plan with an external tracker file, context-clearing checkpoints, and file-state verification | Any multi-phase Claude Code build that won't fit one session — you want zero compaction drift, tollbooths per phase, and installs verified against physical file state. |
| replicate | Replicate-Anything Capability Transfer | A portable capability package (blueprint guide + artifacts + repo command center) enabling one-shot repeatability | You want to transfer a frontend AI capability, a new API, or a playground scaffold into a repeatable Claude Code system via failure-inclusive blueprints and documentation injection. |
Quick Reference
Execution Prompts (structure-pure v2)
4 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 — Human Tollbooth Developer —
skills/mark-kashef-agent-orchestration/references/prompts-v2/human-tollbooth-developer.md
- Mark Kashef — Hybrid Architect Pipeline —
skills/mark-kashef-agent-orchestration/references/prompts-v2/hybrid-architect.md
- Mark Kashef — Parallel Synthesis Advisor —
skills/mark-kashef-agent-orchestration/references/prompts-v2/parallel-synthesis-advisor.md
- Mark Kashef — Sequential Assembly Orchestrator —
skills/mark-kashef-agent-orchestration/references/prompts-v2/sequential-assembly-orchestrator.md
