📊 Analyze Command

The analyze command estimates token usage and cost for a given prompt.

---

🧱 Usage

promptstream analyze --template "<path or text>" [--var key=value ...] [--model <modelId>] [--output json|table]

Options

Option	Description
--template	Inline text or path to a .txt or .json template. (Required)
--var	One or more variables in key=value format.
--model	Optional model ID (defaults to gpt-4o-mini).
--output	Output format: table (default) or json.

---

🧪 Examples

Analyze a file template

promptstream analyze --template "./examples/summary.txt" --var topic=AI --model gpt-4o-mini

Output

📊 Prompt Analysis
────────────────────────────────────────
Model: gpt-4o-mini
Prompt Tokens:     82
Completion Tokens: 0
Total Tokens:      82
Estimated Cost:    $0.000000
────────────────────────────────────────

Output as JSON

promptstream analyze --template "./examples/summary.txt" --var topic=AI --output json

Output

{
  "PromptTokens": 82,
  "CompletionTokens": 0,
  "TotalTokens": 82,
  "EstimatedCost": 0.000000,
  "TimestampUtc": "2025-10-12T12:00:00Z"
}

---

💡 Notes

• The analyze command uses shared models from Flow.AI.Core and the token estimation logic in PromptStream.AI.Services.PromptStreamService.
• Ideal for estimating cost before sending a request to your model.
• Use --output json to pipe structured data into other tools or workflows.