Skip to content

prompt_rewriter.md

Latest commit

 

History

History
154 lines (122 loc) · 5.08 KB

prompt_rewriter.md

File metadata and controls

154 lines (122 loc) · 5.08 KB
Raw

Prompt Token Rewriter

Domain: optimization Skill ID: optimization/prompt_rewriter Issuer: @rosspeili (@ARPAHLS)

Skill Library · Testing

A powerful middleware skill that acts as a deterministic compression logic gate for agents. It ingests a massive, bloated prompt or conversation history and "rewrites" it to use fewer tokens while aggressively retaining 100% of the semantic meaning and instructions.

This is critical for complex agents facing strict token constraints or high LLM API costs.

Manifest Details

Parameters Schema:

  • raw_text (string): The bloated, repetitive prompt or extensive conversation history to compress.
  • compression_aggression (string): The level of compression: 'low', 'medium', or 'high'.

Outputs Schema:

  • compressed_text (string): The aggressively shortened prompt retaining semantic constraints.
  • original_tokens (integer): The approximate original length.
  • new_tokens (integer): The approximate new length.
  • tokens_saved (integer): The absolute number of tokens removed.

Usage Examples

Guides: Usage index · Agent loops. No skill-specific API keys.

Sample user message: Compress this prompt before the main model call: "Hello, could you please make sure to read this documentation..."

Direct execute

from skillware.core.loader import SkillLoader

bundle = SkillLoader.load_skill("optimization/prompt_rewriter")
rewriter = bundle["module"].PromptRewriter()
result = rewriter.execute({
    "raw_text": "Hello, could you please make sure to read this documentation...",
    "compression_aggression": "high",
})
print(result["compressed_text"])

Gemini

import os
import google.genai as genai
from google.genai import types
from skillware.core.env import load_env_file
from skillware.core.loader import SkillLoader

load_env_file()
bundle = SkillLoader.load_skill("optimization/prompt_rewriter")
skill = bundle["module"].PromptRewriter()
client = genai.Client()
tool = SkillLoader.to_gemini_tool(bundle)
response = client.models.generate_content(
    model="gemini-2.5-flash",
    contents="Rewrite this support prompt for a concise, policy-safe assistant.",
    config=types.GenerateContentConfig(
        tools=[tool],
        system_instruction=bundle["instructions"],
    ),
)
for part in response.candidates[0].content.parts:
    if part.function_call:
        result = skill.execute(dict(part.function_call.args))
        follow_up = client.models.generate_content(
            model="gemini-2.5-flash",
            contents=[
                "Use this tool result to answer the original request.",
                {
                    "function_response": {
                        "name": part.function_call.name,
                        "response": {"result": result},
                    }
                },
            ],
            config=types.GenerateContentConfig(
                tools=[tool],
                system_instruction=bundle["instructions"],
            ),
        )
        print(follow_up.text)

Claude

import os
import anthropic
from skillware.core.env import load_env_file
from skillware.core.loader import SkillLoader

load_env_file()
bundle = SkillLoader.load_skill("optimization/prompt_rewriter")
skill = bundle["module"].PromptRewriter()
client = anthropic.Anthropic(api_key=os.environ.get("ANTHROPIC_API_KEY"))
tools = [SkillLoader.to_claude_tool(bundle)]
# On tool_use (name optimization/prompt_rewriter): skill.execute(tool_use.input)

OpenAI

import os
from openai import OpenAI
from skillware.core.env import load_env_file
from skillware.core.loader import SkillLoader

load_env_file()
bundle = SkillLoader.load_skill("optimization/prompt_rewriter")
skill = bundle["module"].PromptRewriter()
openai_tool = SkillLoader.to_openai_tool(bundle)
client = OpenAI(api_key=os.environ.get("OPENAI_API_KEY"))
# Match tool_call.function.name to openai_tool["function"]["name"] (optimization_prompt_rewriter)

DeepSeek

import os
from openai import OpenAI
from skillware.core.env import load_env_file
from skillware.core.loader import SkillLoader

load_env_file()
bundle = SkillLoader.load_skill("optimization/prompt_rewriter")
skill = bundle["module"].PromptRewriter()
deepseek_tool = SkillLoader.to_deepseek_tool(bundle)
client = OpenAI(
    api_key=os.environ.get("DEEPSEEK_API_KEY"),
    base_url="https://api.deepseek.com",
)

Ollama

SkillLoader.to_ollama_prompt(bundle); match "tool": "optimization/prompt_rewriter". See Ollama usage.

Maintenance

To run tests specifically for this skill:

pytest tests/skills/optimization/test_prompt_rewriter.py

Enterprise disclaimer

This skill is provided for demonstration and integration purposes. It is intended as a starting point that you can adapt to your own data, schemas, and operational requirements. For an enterprise-grade version of this skill with dedicated support, SLAs, and customization, contact skills@arpacorp.net.