HTML Prompt Response Construction

HPRC Documentation

An AI-native server-side templating engine that integrates with any web framework — FastAPI, Flask, Django — or runs standalone. Embed LLM prompts directly inside your HTML and have them executed during rendering — like JSP or PHP, but for prompts.

pip install hprc-framework        # then: import hprc

What HPRC does

You write a normal HTML page and drop <prompt> blocks into it. At render time, HPRC resolves each prompt's text (filling in data, request parameters and the responses of other prompts), figures out the execution order automatically, runs prompts concurrently where it can, caches results, and stitches each <response> back into the final HTML. Prompts themselves are tacit — they never appear in the output.

Thirty-second taste

<h1>Hello <fill>customer.name</fill></h1>

<prompt id="summary" model="gpt-5" condition="is_premium" cache="24h">
  Summarize the account for <fill>customer.name</fill>
  who is interested in <param>product</param>.
</prompt>

<section><response id="summary"/></section>

import hprc
from hprc import HPRCConfig, MockLLMClient

config = HPRCConfig(
    llm_client=MockLLMClient(),
    rules={"is_premium": lambda ctx: ctx["customer"]["tier"] == "premium"},
)

html = await hprc.render_template(
    template_path="customer.sprep.html",
    request={"query": {"product": "WidgetPro"}, "path": {}, "method": "GET"},
    bindings={"customer": {"name": "Ada", "tier": "premium"}},
    config=config,
)

The application developer writes no prompt-orchestration logic. See the User Guide for the full walkthrough.

Status