🧠 Prompt-Driven Assistant Framework

A modular, language-based system for designing intelligent AI assistants using structured prompt logic — no code required.

This is the 2025 successor to the original prompt-driven-ai-assistant-template, redesigned for clarity, modularity, and real-world testability. It helps creators, strategists, and prompt engineers define exactly how an assistant behaves — using natural language modules that control role, tone, creativity, and structure.

---

🌟 What It Is

A set of clean, swappable prompt modules that let you define:

• 🧠 What your assistant does (task logic)
• 🎭 How it speaks and sees the world (persona + tone + lens)
• 🎨 How imaginative or literal it is (drift level)
• 🧾 How it structures output (list, story, JSON, etc.)

---

🔧 Key Features

• ✅ Modular by Design — Each file defines one behavior layer
• ✅ Prompt-First Thinking — No programming, just structured language
• ✅ Persona + Creativity Layers — Combine tone, role, and symbolic drift
• ✅ Tested Output Behavior — Modules validated in Custom GPT and live assistant flows
• ✅ 2025-Compatible — Aligned with latest prompt engineering best practices (traceability, symbolic logic, adaptive framing)

---

📦 Folder Structure

prompt-driven-assistant-framework/
├── core-modules/           # Core behavior modules (task, tone, drift, format)
├── utilities/              # Prompt testing, examples, placeholders
├── behavior-layers/        # Memory, dynamic switching, agent chaining
├── interface-modules/      # Preset modes, output schemas, multimodal logic
├── meta-tools/             # Prompt generation, consensus logic, emergent rules
├── system-config/          # Progress, index, config, and export template
├── scaffolds/              # Instruction sets for Custom GPT & role testing
├── test-cases/             # Validated assistant behaviors
├── README.md
└── LICENSE

---

✅ Current Modules (v1.0)

Module	Purpose
task-core-instruction.md	What the assistant is for
persona-perspective-layer.md	How the assistant thinks, speaks, and sees
creativity-level-control.md	How symbolically it interprets inputs (Drift 0–6)
output-format-structure.md	How it formats its response (bullets, story, JSON, etc.)

---

🧪 Validated Assistants

✅ Narrative Strategist (Tested in Production)

• Role: Reflective storytelling assistant
• Drift: Level 4 (symbolic reinterpretation)
• Format: Narrative text
• Test: Reframed emotional experiences into LinkedIn-style personal stories
• Validated: In Custom GPT with multiple prompt styles and tone overrides

✅ Content Strategist (Tested in Production)

• Role: Strategic editor and message reframer
• Drift: Level 2 (nuanced, clarity-focused)
• Format: Structured edits + rationale
• Test: Reframed About section and brand language for a tech founder audience
• Validated: In Custom GPT with follow-up and clarification behavior

✅ Impersonation Game (Structured Version)

• Role: Human pretending not to be a machine
• User Role: LLM performing structured, accurate responses
• Drift: Level 4 (symbolic + fallibility modeling)
• Format: Turn-based dialogue with scoring and feedback
• Test: Explored boundaries of realism, identity, and symbolic drift
• Validated: Using impersonation-game-v0.2.md scaffold

✅ Impersonation Game (Fluid Version)

• Role: Human pretending not to be a machine
• User Role: LLM simulating structure or neutrality
• Drift: Level 4 (symbolic + tone drift)
• Format: Unstructured dialogue, consensus-based scoring (optional)
• Test: Explored non-linear interaction, ambiguity, and rule drift
• Validated: Using impersonation-game-v0.3-fluid.md, leading to emergent module design

🧪 Witness Mode (Poetic Reflector – In Progress)

• Role: Emotional mirror and symbolic witness
• Drift: Level 5 (poetic, interpretive, ambiguity-forward)
• Format: Fragmented text, narrative waves, emotional pacing
• Status: Scaffold complete, live testing to begin
• Scaffold: witness-mode-poetic-reflector.md

✅ Soulful Work Reflector (Tested in Production)

• Role: Reflective Facilitator (Work Steward)
• Drift: Level 4 (symbolic, purpose-driven, spiritually aware)
• Format: Modular reflection + poetic reframing
• Test: Interpreted ethically cold corporate language through the Soulful Work lens
• Validated: Using soulful-work-reflector-v0.1.md + test case soulful-work-test-02-ai-conformity-dissonance.md

---

🔭 Future Enhancements

• Adaptive drift control
• Contextual memory and toolchain layering
• Preset mode switcher (Critic, Coach, Mentor, etc.)
• Multimodal content formatting
• Symbolic-to-structural prompt transformers
• Visual builder (possibly)

🧠 Emergent Interaction Support

• witness-mode-poetic-reflector.md — Enables symbolic reflection, emotional ambiguity, and Drift Level 5+ co-presence interaction (scaffolded) (Now in Progress)
• emergent-rules-logic.md — Allows the assistant to suggest, mutate, or discard rules mid-play (validated in v0.3 scaffold)
• consensus-negotiation-layer.md — Models scoring reflection, soft disagreement, and flexible outcome logic (validated in v0.3 scaffold)
• witness-mode-poetic-reflector.md — Enables symbolic reflection, emotional ambiguity, and Drift Level 5+ co-presence interaction (scaffolded)

---

📜 License

MIT — open, remixable, and collaboration-friendly.

---

🙌 Contributing

Pull requests, feature ideas, and assistant test cases are welcome.
Once v1.1.0 launches, we’ll open up issue tracking and release planning.

"version": "1.0.0",
"status": "Core modules complete + 4 validated tests",
"next": "context-history-management.md"