PromptStream.AI 0.1.5

<p align="center"> <img src="https://github.com/AndrewClements84/PromptStream.AI/blob/master/assets/logo.png?raw=true" alt="PromptStream.AI" width="500"/> </p>

🧠 PromptStream.AI

🧩 Description

PromptStream.AI — Token-aware prompt composition, validation, and conversational context toolkit for .NET.

Built atop Flow.AI.Core and TokenFlow.AI,
PromptStream.AI enables developers to compose, validate, generate, and manage multi-turn AI prompts with token budgeting, interpolation, and contextual memory.

⚙️ Key Features

• 🧩 Token-aware prompt builder with variable interpolation ({{variable}} syntax)
• ✅ Validation engine for token limits, structure, and completeness
• 🔢 Integrated token counting via ITokenFlowProvider
• 🧠 Context manager with replay, merge, summarization, and JSON persistence
• 💬 Assistant reply tracking and multi-turn chat history
• 💾 Persistent context storage (ToJson / LoadFromJson)
• 🧮 Token budgeting tools (EstimateTokenUsage, TrimToTokenBudget)
• ⚡ CLI utility (PromptStream.AI.CLI) for building, validating, and generating prompts
• 🔌 Seamless integration with TokenFlow.AI for model-aware tokenization

🚀 Installation

dotnet add package PromptStream.AI

Requires:

• .NET 8.0 or higher
• Flow.AI.Core
• (optional) TokenFlow.AI for model-specific token counting

🧠 Quickstart Example

using System;
using System.Collections.Generic;
using TokenFlow.AI.Integration;
using PromptStream.AI.Models;
using PromptStream.AI.Services;

// Create a token provider (TokenFlow.AI adapter)
var tokenProvider = new TokenFlowProvider("gpt-4o-mini");

// Initialize PromptStream service (with context-aware memory)
var service = new PromptStreamService(tokenProvider);

// Define a prompt template
var template = new PromptTemplate
{
    Id = "summarize-v1",
    Template = "Summarize the following:\n\n{{input}}\n\nBe concise.",
    RequiredVariables = new() { "input" }
};

// Build and validate
var variables = new Dictionary<string, string>
{
    ["input"] = "Flow.AI enables composable AI workflows for .NET developers."
};

var (instance, validation) = service.BuildAndValidate(template, variables);

if (validation.IsValid)
{
    Console.WriteLine($"✅ Valid prompt ({validation.TokenCount} tokens)");
    Console.WriteLine(instance.RenderedText);
}
else
{
    Console.WriteLine($"❌ Invalid: {string.Join(", ", validation.Errors)}");
}

// Add an assistant reply for multi-turn context
service.AddAssistantReply("Sure! Here's a short summary...");

💻 CLI Usage (PromptStream.AI.CLI)

PromptStream.AI now includes a lightweight command-line interface for developers to build, validate, and generate prompts directly from the terminal.

🧩 Build a prompt

dotnet run --project src/PromptStream.AI.CLI/PromptStream.AI.CLI.csproj -- build --template "Hello {{name}}" --var name=Andrew

✅ Validate a prompt

dotnet run --project src/PromptStream.AI.CLI/PromptStream.AI.CLI.csproj -- validate --template "Summarize {{topic}}" --var topic="AI in .NET"

🤖 Generate a model response

dotnet run --project src/PromptStream.AI.CLI/PromptStream.AI.CLI.csproj -- generate --template "Explain {{concept}}" --var concept="tokenization" --save context.json

🧠 Manage conversation context

dotnet run --project src/PromptStream.AI.CLI/PromptStream.AI.CLI.csproj -- context --load context.json --summarize

Available commands: | Command | Description | |----------|--------------| | build | Render a prompt with variable substitution | | validate | Validate prompt completeness and token limits | | generate | Build, validate, and produce a model-like response | | context | Load, save, summarize, or clear conversation context |

🧩 Namespace Overview

🧪 Testing & Coverage

dotnet test --collect:"XPlat Code Coverage"

All core logic is tested using xUnit and coverlet, with full integration into Codecov for continuous coverage tracking.

🗺️ Roadmap

🤝 Contributing

Contributions are welcome!
If you’d like to extend functionality or improve coverage, please fork the repo and submit a pull request.

See CONTRIBUTING.md for guidelines.

📜 License

This project is licensed under the MIT License — see the LICENSE file for details.

Part of the Flow.AI Ecosystem
© 2025 Andrew Clements