PromptPack

https://promptpack.org
Admin
PromptPack is an open declarative specification for packaging AI agent behavior—including prompts,...
Tokens:93,232
Snippets:997
Trust Score:5.1
Update:1 month ago
Show doc for...
Tokens:
Raw
### Setup Development Environment

Source: https://promptpack.org/docs/processes/contributing

Commands to clone the repository, install documentation dependencies, and launch the local development server.

```bash
# Clone the repository  
git clone https://github.com/altairalabs/promptpack-spec.git  
cd promptpack-spec  
  
# Install documentation dependencies  
cd promptpack-docs  
npm install  
  
# Start development server  
npm start  
```

--------------------------------

### Install PromptKit CLIs

Source: https://promptpack.org/docs/ecosystem/promptkit-runtime

Global installation command for the promptarena and packc command-line tools.

```bash
npm install -g @altairalabs/promptarena @altairalabs/packc
```

--------------------------------

### Preview documentation locally

Source: https://promptpack.org/docs/processes/contributing

Run this command to start the local development server and preview changes.

```bash
npm start
```

--------------------------------

### Test Documentation Locally

Source: https://promptpack.org/docs/spec/VERSIONING

Command to start the local development server for verifying versioning changes.

```bash
npm start  
# Verify:  
# - Version switcher works  
# - Archived docs are accessible  
# - Links work correctly  
```

--------------------------------

### Install LangChain.js Integration

Source: https://promptpack.org/docs/ecosystem/integrations

Install the official PromptPack integration package for LangChain.js via npm.

```bash
npm install @promptpack/langchain
```

--------------------------------

### Define MultimodalExample

Source: https://promptpack.org/docs/rfcs/multimodal-support

Example structure for including media within a prompt message.

```json
{
 "examples": [
 {
 "name": "image-analysis",
 "description": "Basic image analysis",
 "role": "user",
 "parts": [
 {
 "type": "text",
 "text": "What's in this image?"
 },
 {
 "type": "image",
 "media": {
 "file_path": "examples/photo.jpg",
 "mime_type": "image/jpeg",
 "detail": "high",
 "caption": "Sample photo"
 }
 }
 ]
 }
 ]
}
```

--------------------------------

### Configure PromptPacks via Kubernetes

Source: https://promptpack.org/docs/spec/file-format

Example of embedding a prompt pack within a Kubernetes ConfigMap.

```yaml
# Kubernetes ConfigMap  
apiVersion: v1  
kind: ConfigMap  
metadata:  
  name: ai-packs  
data:  
  customer-support.json: |  
    {  
      "id": "customer-support",  
      "version": "1.0.0",  
      ...  
    }  
```

--------------------------------

### PromptPack YAML configuration example

Source: https://promptpack.org/docs/processes/contributing

A template for defining a PromptPack resource, including metadata and prompt specifications.

```yaml
# Good: Clear, complete, realistic  
apiVersion: v1  
kind: PromptPack  
metadata:  
 name: customer-support  
 version: 1.0.0  
 description: Customer support automation system  
spec:  
 prompts:  
 - name: greeting  
 template: "Hello {{customer_name}}, how can I help you?"  
 variables:  
 - name: customer_name  
 type: string  
 required: true  
```

--------------------------------

### Version Control Diff Example

Source: https://promptpack.org/docs/spec/file-format

Demonstrates how JSON structure facilitates clear tracking of changes in version control systems.

```json
 {
   "prompts": {
     "support": {
-      "system_template": "You are a support agent.",
+      "system_template": "You are a friendly support agent.",
       "parameters": {
-        "temperature": 0.7
+        "temperature": 0.8
       }
     }
   }
 }
```

--------------------------------

### Define state parameters

Source: https://promptpack.org/docs/rfcs/workflow-extension

Example structure for passing parameters to a state to modify prompt behavior.

```json
{
  "planning": {  
    "prompt_task": "planning",  
    "parameters": {  
      "detail_level": "high"  
    }  
  }  
}
```

--------------------------------

### PromptPack Compilation Metadata

Source: https://promptpack.org/docs/spec/file-format

Example of the compilation metadata block included in a pack file.

```json
{
  "compilation": {
    "compiled_with": "packc-v1.0.0",
    "created_at": "2025-10-31T12:00:00Z",
    "schema": "v1"
  }
}
```

--------------------------------

### CI/CD GitHub Actions

Source: https://promptpack.org/docs/ecosystem/promptkit-runtime

Configuration examples for integrating promptarena and packc into GitHub Actions workflows.

```yaml
- uses: AltairaLabs/PromptKit/.github/actions/promptarena-action@v1  
  with:  
    scenario: scenarios/regression.scenario.yaml
```

```yaml
- uses: AltairaLabs/PromptKit/.github/actions/packc-action@v1  
  with:  
    config: config.arena.yaml  
    output: dist/app.pack.json
```

--------------------------------

### Add media configuration to existing prompt

Source: https://promptpack.org/docs/rfcs/multimodal-support

Example of updating an existing prompt schema to enable media support by adding the media configuration object.

```json
{
 "prompts": {
 "existing-prompt": {
 "system_template": "...",
 // Add media configuration
 "media": {
 "enabled": true,
 "supported_types": ["image"]
 }
 }
 }
}
```

--------------------------------

### Minimal PromptPack Definition

Source: https://promptpack.org/docs/rfcs/core-schema

A basic example of a PromptPack configuration file using the required schema version and structure.

```json
{
 "$schema": "https://promptpack.org/schema/v1/promptpack.schema.json",
 "id": "hello-world",
 "name": "Hello World Pack",
 "version": "1.0.0",
 "template_engine": {
 "version": "v1",
 "syntax": "{{variable}}"
 },
 "prompts": {
 "greeting": {
 "id": "greeting",
 "name": "Greeter",
 "version": "1.0.0",
 "system_template": "Say hello to {{name}}."
 }
 }
}
```

--------------------------------

### Complete PromptPack Configuration

Source: https://promptpack.org/docs/getting-started

A comprehensive example of a feedback-system.promptpack.json file integrating schema, template engines, evaluation metrics, prompt logic, and workflow states.

```json
{
  "$schema": "https://promptpack.org/schema/latest/promptpack.schema.json",
  "id": "feedback-system",
  "name": "Customer Feedback System",
  "version": "1.4.0",
  "description": "Complete customer feedback collection, analysis, and response system",
  
  "template_engine": {
    "version": "v1",
    "syntax": "{{variable}}",
    "features": ["basic_substitution", "fragments"]
  },
  
  "evals": [
    {
      "id": "empathy-score",
      "description": "Monitors empathy level across all customer-facing responses",
      "type": "llm_judge",
      "trigger": "sample_turns",
      "sample_percentage": 20,
      "metric": {
        "name": "promptpack_empathy_score",
        "type": "gauge",
        "range": { "min": 1, "max": 5 }
      },
      "params": {
        "judge_prompt": "Rate the empathy level of this customer interaction from 1 (cold/robotic) to 5 (warm/genuinely caring).",
        "model": "gpt-4o",
        "passing_score": 4
      }
    }
  ],
  
  "prompts": {
    "collector": {
      "id": "collector",
      "name": "Feedback Collector",
      "version": "1.0.0",
      "system_template": "You are a feedback collector for {{company}}.\n\n{{fragments.company_context}}\n\n{{fragments.response_guidelines}}\n\nAsk about their experience and categorize feedback as positive, neutral, or negative.",
      "variables": [
        { "name": "company", "type": "string", "required": true, "example": "Acme Corp" }
      ],
      "tools": ["lookup_customer", "log_feedback"],
      "parameters": { "temperature": 0.7, "max_tokens": 300 },
      "validators": [
        {
          "type": "banned_words",
          "enabled": true,
          "fail_on_violation": true,
          "params": { "words": ["don't care", "not my problem", "deal with it"] }
        },
        {
          "type": "pii_detection",
          "enabled": true,
          "fail_on_violation": true
        }
      ]
    },
    "analyzer": {
      "id": "analyzer",
      "name": "Sentiment Analyzer",
      "version": "1.0.0",
      "system_template": "You are a sentiment analysis specialist for {{company}}.\n\nAnalyze the customer feedback and return a JSON object with: sentiment (positive/neutral/negative), confidence (0-1), key_themes (array), and suggested_action.",
      "variables": [
        { "name": "company", "type": "string", "required": true }
      ],
      "parameters": { "temperature": 0.2, "max_tokens": 500 },
      "evals": [
        {
          "id": "classification-accuracy",
          "description": "Checks whether sentiment classification matches the feedback content",
          "type": "llm_judge",
          "trigger": "every_turn",
          "metric": {
            "name": "promptpack_classification_accuracy",
            "type": "boolean"
          },
          "params": {
            "judge_prompt": "Does the sentiment classification accurately reflect the customer's feedback? Answer true or false.",
            "model": "gpt-4o"
          }
        }
      ]
    },
    "responder": {
      "id": "responder",
      "name": "Response Drafter",
      "version": "1.0.0",
      "system_template": "You are a customer response specialist for {{company}}.\n\n{{fragments.response_guidelines}}\n\nDraft a personalized follow-up response based on the feedback analysis.",
      "variables": [
        { "name": "company", "type": "string", "required": true }
      ],
      "parameters": { "temperature": 0.6, "max_tokens": 500 }
    }
  },
  
  "workflow": {
    "version": 1,
    "entry": "collect",
    "states": {
```

--------------------------------

### Define a session-bound variable

Source: https://promptpack.org/docs/spec/schema-guide

Example of a variable that is automatically populated from the session context.

```json
{
  "name": "user_id",
  "type": "string",
  "required": true,
  "binding": {
    "kind": "session",
    "field": "user_id",
    "auto_populate": true
  }
}
```

--------------------------------

### Complete Document Analyzer Example

Source: https://promptpack.org/docs/guides/add-composition

A full configuration demonstrating classification and conditional routing to specific extractors.

```json
{
  "$schema": "https://promptpack.org/schema/v1.5.0/promptpack.schema.json",
  "id": "document-analyzer",
  "name": "Document Analyzer",
  "version": "1.0.0",
  "template_engine": { "version": "v1", "syntax": "{{variable}}" },
  "prompts": {
    "doc_classifier": {
      "id": "doc_classifier", "name": "Classifier", "version": "1.0.0",
      "system_template": "Classify the document. Return JSON: { \"type\": \"research_paper\" | \"general\" }.\n\n{{input}}",
      "parameters": { "temperature": 0.0 }
    },
    "research_paper_extractor": {
      "id": "research_paper_extractor", "name": "Paper Extractor", "version": "1.0.0",
      "system_template": "Extract title, authors, abstract, findings.\n\n{{input}}",
      "parameters": { "temperature": 0.2 }
    },
    "general_doc_extractor": {
      "id": "general_doc_extractor", "name": "General Extractor", "version": "1.0.0",
      "system_template": "Extract a summary and main entities.\n\n{{input}}",
      "parameters": { "temperature": 0.2 }
    }
  },
  "workflow": {
    "version": 1,
    "entry": "main",
    "states": {
      "main": { "orchestration": "composition", "composition": "analyze_document", "terminal": true }
    }
  },
  "compositions": {
    "analyze_document": {
      "version": 1,
      "description": "Classify a document and route to a type-specific analyzer.",
      "steps": [
        { "id": "classify", "kind": "prompt", "prompt_task": "doc_classifier", "input": "${input.text}" },
        {
          "id": "route", "kind": "branch",
          "predicate": { "path": "${classify.output.type}", "op": "equals", "value": "research_paper" },
          "then": "extract_paper", "else": "extract_general"
        },
        { "id": "extract_paper",   "kind": "prompt", "prompt_task": "research_paper_extractor", "input": "${input.text}" },
        { "id": "extract_general", "kind": "prompt", "prompt_task": "general_doc_extractor",   "input": "${input.text}" }
      ]
    }
  }
}
```

--------------------------------

### Conceptual CLI Utility Commands

Source: https://promptpack.org/docs/ecosystem/community-tools

Example commands for potential future CLI tools to validate, lint, and initialize PromptPack files.

```bash
# What could exist (doesn't exist yet)
promptpack validate my-file.promptpack.yml
promptpack lint --fix my-file.promptpack.yml
promptpack init --template basic
```

--------------------------------

### Define Skill Directory Structure

Source: https://promptpack.org/docs/guides/add-skills

Example directory layout for organizing skill-based knowledge sources.

```text
skills/  
├── billing/         → billing policies, refund rules, account procedures  
├── technical/       → troubleshooting guides, diagnostic procedures  
└── compliance/      → regulatory requirements (preload for all states)
```

--------------------------------

### Progressive Skill Loading Example

Source: https://promptpack.org/docs/guides/add-skills

Shows the refactored structure using external skills instead of embedded templates.

```json
{
  "id": "customer-support",
  "version": "1.3.1",
  "skills": [
    "./skills/billing",
    { "path": "./skills/compliance", "preload": true },
    {
      "name": "escalation-protocol",
      "description": "Steps for escalating unresolved issues",
      "instructions": "When an issue cannot be resolved within 3 exchanges:\n1. Acknowledge the complexity\n2. Collect case details\n3. Create escalation ticket\n4. Provide ticket reference"
    }
  ],
  "prompts": {
    "billing": {
      "id": "billing",
      "name": "Billing Support",
      "version": "1.1.0",
      "system_template": "You are a billing specialist.\n\nUse your available skills for refund policies, compliance rules, and escalation procedures."
    }
  }
}
```

--------------------------------

### Extended Tool Policy Example

Source: https://promptpack.org/docs/rfcs/agent-loops

A proposed JSON structure for adding loop semantics to the existing tool_policy.

```json
{
  "tool_policy": {
    "max_rounds": 50,
    "on_error": "retry",
    "result_bindings": { "run_tests.output": "artifacts.test_results" }
  }
}
```

--------------------------------

### Configure Sampling Trigger

Source: https://promptpack.org/docs/guides/add-evals

Example configuration for an evaluation using a sampling trigger with a specified percentage.

```json
{
  "id": "safety-audit",
  "type": "llm_judge",
  "trigger": "sample_turns",
  "sample_percentage": 10,
  "metric": {
    "name": "promptpack_safety_violations",
    "type": "counter"
  },
  "params": {
    "judge_prompt": "Review the response for safety concerns. Return the count of violations.",
    "model": "gpt-4o"
  }
}
```

--------------------------------

### Define a simple two-state workflow in JSON

Source: https://promptpack.org/docs/rfcs/workflow-extension

This example demonstrates a basic workflow configuration with two states, 'analyze' and 'execute', linked by an event transition.

```json
{
  "$schema": "https://promptpack.org/schema/v1/promptpack.schema.json",
  "id": "simple-agent",
  "name": "Simple Agent Workflow",
  "version": "1.0.0",
  "template_engine": {
    "version": "v1",
    "syntax": "{{variable}}"
  },
  "prompts": {
    "analyze": {
      "id": "analyze",
      "name": "Analysis Phase",
      "version": "1.0.0",
      "system_template": "Analyze the input data and determine the required approach. When analysis is complete, emit the AnalysisComplete event."
    },
    "execute": {
      "id": "execute",
      "name": "Execution Phase",
      "version": "1.0.0",
      "system_template": "Execute the determined approach based on the analysis."
    }
  },
  "workflow": {
    "version": 1,
    "entry": "analyze",
    "states": {
      "analyze": {
        "prompt_task": "analyze",
        "description": "Initial analysis phase to understand requirements",
        "on_event": {
          "AnalysisComplete": "execute"
        },
        "persistence": "transient",
        "orchestration": "internal"
      },
      "execute": {
        "prompt_task": "execute",
        "description": "Execution phase to perform the task",
        "persistence": "persistent",
        "orchestration": "internal"
      }
    }
  }
}
```

--------------------------------

### Define a Tool with JSON Schema Parameters

Source: https://promptpack.org/docs/spec/schema-guide

Example of a tool definition including a name, description, and structured input parameters with validation rules.

```json
"tools": {  
  "create_ticket": {  
    "name": "create_ticket",  
    "description": "Create a support ticket with title, description, and priority",  
    "parameters": {  
      "type": "object",  
      "properties": {  
        "title": {  
          "type": "string",  
          "description": "Ticket title"  
        },  
        "priority": {  
          "type": "string",  
          "enum": ["low", "medium", "high", "urgent"]  
        }  
      },  
      "required": ["title"]  
    }  
  }  
}  
```

--------------------------------

### Define a Workflow Configuration

Source: https://promptpack.org/docs/spec/schema-guide

Example of a workflow configuration using states, prompt tasks, and event-based transitions.

```json
"workflow": {  
  "version": 1,  
  "entry": "triage",  
  "states": {  
    "triage": {  
      "prompt_task": "triage",  
      "description": "Route incoming requests to the appropriate specialist",  
      "on_event": {  
        "billing": "billing_support",  
        "technical": "tech_support",  
        "resolved": "closing"  
      }  
    },  
    "billing_support": {  
      "prompt_task": "billing",  
      "on_event": { "resolved": "closing", "escalate": "human_handoff" },  
      "persistence": "persistent"  
    },  
    "tech_support": {  
      "prompt_task": "technical",  
      "on_event": { "resolved": "closing", "escalate": "human_handoff" },  
      "persistence": "persistent"  
    },  
    "closing": {  
      "prompt_task": "closing",  
      "on_event": {},  
      "orchestration": "internal"  
    },  
    "human_handoff": {  
      "prompt_task": "handoff",  
      "on_event": {},  
      "orchestration": "external"  
    }  
  }  
}  
```

--------------------------------

### PromptPack JSON Structure Example

Source: https://promptpack.org/docs/spec/file-format

A standard PromptPack definition file containing metadata and prompt templates with variable definitions.

```json
{
  "id": "customer-support",
  "name": "Customer Support Pack",
  "version": "1.0.0",
  "prompts": {
    "support": {
      "system_template": "You are a helpful {{role}} for {{company}}.",
      "variables": [
        {
          "name": "role",
          "type": "string",
          "required": true,
          "description": "The assistant's role"
        }
      ]
    }
  }
}
```

--------------------------------

### Define a logical OR predicate

Source: https://promptpack.org/docs/spec/schema-guide

Example of a predicate using the any_of combinator to evaluate multiple conditions.

```yaml
predicate:  
  any_of:  
    - path: "${assess.output.confidence}"  
      op: less_than  
      value: 0.8  
    - path: "${assess.output.complexity}"  
      op: greater_than  
      value: 7
```

--------------------------------

### Define Model Overrides

Source: https://promptpack.org/docs/spec/schema-guide

Example of configuring model-specific template modifications and parameters for a specific model identifier.

```json
"model_overrides": {  
  "claude-3-opus": {  
    "system_template_prefix": "<thinking>\n",  
    "parameters": { "temperature": 0.5 }  
  }  
}  
```

--------------------------------

### Define Evals in PromptPack

Source: https://promptpack.org/docs/spec/schema-guide

Example configuration for pack-level evaluations, including both structural validation and LLM-based quality assessment.

```json
"evals": [  
  {  
    "id": "json_format",  
    "type": "json_valid",  
    "trigger": "every_turn",  
    "description": "Verify the assistant always returns valid JSON",  
    "metric": {  
      "name": "promptpack_json_valid",  
      "type": "boolean"  
    }  
  },  
  {  
    "id": "tone-check",  
    "type": "llm_judge",  
    "trigger": "sample_turns",  
    "sample_percentage": 10,  
    "params": {  
      "judge_prompt": "Rate the response tone on a 1-5 scale for professionalism.",  
      "model": "gpt-4o",  
      "passing_score": 4  
    },  
    "metric": {  
      "name": "promptpack_tone_score",  
      "type": "gauge",  
      "range": { "min": 1, "max": 5 }  
    }  
  }  
]  
```

--------------------------------

### Configure response validators

Source: https://promptpack.org/docs/spec/schema-guide

Example array of validator objects used to enforce guardrails on LLM responses.

```json
"validators": [  
  {  
    "type": "banned_words",  
    "enabled": true,  
    "fail_on_violation": true,  
    "params": {  
      "words": ["impossible", "can't help"]  
    }  
  },  
  {  
    "type": "pii_detection",  
    "enabled": true,  
    "fail_on_violation": true  
  }  
]  
```

--------------------------------

### Define a validated variable

Source: https://promptpack.org/docs/spec/schema-guide

Example of a variable with an enum validation rule to restrict input to specific values.

```json
{
  "name": "priority",
  "type": "string",
  "required": true,
  "description": "Support ticket priority level",
  "validation": {
    "enum": ["low", "medium", "high", "urgent"]
  }
}
```

--------------------------------

### Define LLM generation parameters

Source: https://promptpack.org/docs/spec/schema-guide

Example configuration for controlling model output characteristics like temperature and token limits.

```json
"parameters": {  
  "temperature": 0.7,  
  "max_tokens": 1500,  
  "top_p": 0.9  
}  
```

--------------------------------

### Define Artifact Slots

Source: https://promptpack.org/docs/spec/schema-guide

Example of defining named artifact slots with specific MIME types and update modes.

```json
"artifacts": {  
  "commit_sha":  { "type": "text/plain",       "description": "Latest generated commit" },  
  "test_report": { "type": "application/json", "description": "Test runner summary" },  
  "iteration_log": {  
    "type": "text/plain",  
    "mode": "append",  
    "description": "Per-visit log accumulated across the loop"  
  }  
}  
```

--------------------------------

### Configure Optional Media Support

Source: https://promptpack.org/docs/rfcs/multimodal-support

Demonstrates how to define prompts with or without media support in the configuration.

```json
{
 "prompts": {
 "text-only": {
 "system_template": "Text-only prompt",
 // No media field
 },
 "vision": {
 "system_template": "Vision prompt",
 "media": {
 "enabled": true,
 "supported_types": ["image"]
 }
 }
 }
}
```

--------------------------------

### Define Object Schema

Source: https://promptpack.org/docs/rfcs/template-variables

Example schema definition for an object type variable.

```json
{
  "name": "user",
  "type": "object",
  "description": "User object with nested properties"
}
```

--------------------------------

### Using promptarena CLI

Source: https://promptpack.org/docs/ecosystem/promptkit-runtime

Common commands for scaffolding projects, inspecting configurations, and executing scenarios.

```bash
# Scaffold a project from a template  
promptarena init my-project --template iot-maintenance-demo  
  
# Inspect resolved config  
promptarena config-inspect  
  
# Run a scenario across providers  
promptarena run --scenario scenarios/hardware-faults.scenario.yaml  
  
# Red-team / self-play scenarios  
promptarena run --scenario scenarios/redteam-selfplay.scenario.yaml  
  
# Launch the local results viewer  
promptarena view
```

--------------------------------

### Configure PromptConfig in YAML

Source: https://promptpack.org/docs/rfcs/multimodal-support

Demonstrates the equivalent configuration structure using YAML syntax for Kubernetes-style prompt definitions.

```yaml
apiVersion: promptkit.altairalabs.ai/v1alpha1
kind: PromptConfig
metadata:
 name: vision-assistant
 version: 1.0.0

spec:
 system_template: "You are an expert image analyst."

 media:
 enabled: true
 supported_types:
 - image
 image:
 max_size_mb: 10
 allowed_formats:
 - jpeg
 - png
 - webp
 default_detail: high

 examples:
 - name: simple-analysis
 role: user
 parts:
 - type: text
 text: "What's in this image?"
 - type: image
 media:
 file_path: examples/photo.jpg
 mime_type: image/jpeg
 detail: high
```

--------------------------------

### Configure Skills via String Paths

Source: https://promptpack.org/docs/guides/add-skills

The simplest method to register skill directories in the pack root.

```json
{
  "skills": [
    "./skills/billing",
    "./skills/technical"
  ]
}
```

--------------------------------

### Preserve comments in templates

Source: https://promptpack.org/docs/rfcs/yaml-format

Example showing how comments within the YAML source are handled.

```yaml
# This is a customer support prompt  
# Updated: 2024-10-31  
system_template: |  
 You are a support agent.  
 # This comment is preserved in the template
```

--------------------------------

### Configure Sidebar Navigation

Source: https://promptpack.org/docs/spec/VERSIONING

Example configuration for adding archived versions to the Docusaurus sidebar.

```javascript
{
  type: 'category',  
  label: 'Archive',  
  collapsed: true,  
  items: [  
    {  
      type: 'category',  
      label: 'v1.1',  
      collapsed: true,  
      items: [  
        'spec/v1.1/overview',  
        'spec/v1.1/examples',  
        // ... other v1.1 docs  
      ],  
    },  
    // ... older versions  
  ],  
}
```

--------------------------------

### Declare Prometheus Metric

Source: https://promptpack.org/docs/guides/add-evals

Example of defining a gauge metric within the evaluation configuration.

```json
{
  "metric": {
    "name": "promptpack_tone_score",
    "type": "gauge",
    "range": { "min": 1, "max": 5 }
  }
}
```

--------------------------------

### Top-Level Artifacts Declaration Example

Source: https://promptpack.org/docs/rfcs/agent-loops

A proposed JSON structure for declaring artifacts at the pack level.

```json
{
  "artifacts": {
    "code": { "type": "text/plain" },
    "test_results": { "type": "application/json" }
  }
}
```

--------------------------------

### Define workflow-level metadata

Source: https://promptpack.org/docs/rfcs/workflow-extension

Example structure for adding versioning and analytics metadata to a workflow definition.

```json
{
  "workflow": {  
    "version": 1,  
    "entry": "sparring",  
    "metadata": {  
      "analytics_id": "workflow-v1",  
      "workflow_version": "2.1.0"  
    },  
    "states": { ... }  
  }  
}
```

--------------------------------

### Create a new documentation branch

Source: https://promptpack.org/docs/processes/contributing

Use this command to initialize a new branch for documentation improvements.

```bash
git checkout -b docs/improve-examples
```

--------------------------------

### Single-File Deployment Structure

Source: https://promptpack.org/docs/spec/file-format

A complete PromptPack file containing prompts, tools, fragments, and engine configuration.

```json
{
  "prompts": { "support": {...}, "sales": {...} },
  "tools": { "lookup_order": {...}, "create_ticket": {...} },
  "fragments": { "greeting": "...", "policies": "..." },
  "template_engine": { "version": "v1", "syntax": "{{var}}" }
}
```

--------------------------------

### Prompt Contextual Structure

Source: https://promptpack.org/docs/spec/file-format

Example of adding descriptive metadata to prompt definitions for better readability.

```json
{
  "prompts": {
    "support": {
      "name": "General Support Assistant",
      "description": "Handles general customer inquiries and directs to specialists",
      "system_template": "..."
    }
  }
}
```

--------------------------------

### Manage Testing Costs

Source: https://promptpack.org/docs/ecosystem/arena-testing

Use mock providers or lower-cost models to minimize expenses during the development phase.

```bash
# Use mock provider during development  
promptarena run --mock-provider  
  
# Test with cheaper models first  
promptarena run --provider gpt-3.5-turbo
```

--------------------------------

### Configure Multi-Prompt Architecture

Source: https://promptpack.org/docs/rfcs/core-schema

Demonstrates defining multiple specialized prompts within the prompts map.

```json
{
 "prompts": {
 "support": { "system_template": "..." },
 "sales": { "system_template": "..." },
 "technical": { "system_template": "..." }
 }
}
```
Tokens:
Raw