Elsai Prompts

Package: elsai-prompts  v1.1.0

Centralized prompt management with version control and traceability — prompts are stored and managed within Elsai Foundry, giving you a single source of truth for all prompt templates across your projects.

WARNING

v1.0.0 is deprecated. Upgrade to v1.1.0 or later. Prompts have migrated from Pezzo to Elsai Foundry.

Installation

pip install --extra-index-url https://core-packages.elsai.ai/root/elsai-prompts/ elsai-prompts==1.1.0

Requirements: Python >= 3.9

---

Why centralized prompts?

Hard-coding prompts in application code creates drift — different versions scattered across services with no history or rollback. Elsai Prompts gives you:

• Version control — full history of every prompt change
• Environment deployments — separate active versions per environment (dev / staging / prod)
• Traceability — know exactly which prompt version produced which output
• Lifecycle management — draft → review → published → deprecated

---

PromptManager

PromptManager is the main class for connecting to Elsai Foundry and retrieving prompts.

Initialisation

You can provide the API base URL either as a constructor argument or via an environment variable.

Option 1 — pass base_url directly:

from elsai_prompts.prompt_manager import PromptManager

prompt_manager = PromptManager(
    api_key="your_elsai_api_key",
    project_id="your_project_id",
    base_url="https://your-api-url.com",
)

Option 2 — set the environment variable:

export PROMPT_MANAGER_API_URL="https://your-api-url.com"

from elsai_prompts.prompt_manager import PromptManager

prompt_manager = PromptManager(
    api_key="your_elsai_api_key",
    project_id="your_project_id",
)

Constructor parameters:

Parameter	Required	Description
api_key	Yes	Elsai Foundry API key
project_id	Yes	Project ID from your Elsai Foundry workspace
base_url	No	API base URL. If omitted, reads from PROMPT_MANAGER_API_URL env var

---

Fetching a prompt

Use get_active_prompt_version() to retrieve the currently active (published) version of a prompt by name:

from elsai_prompts.prompt_manager import PromptManager

prompt_manager = PromptManager(
    api_key="your_elsai_api_key",
    project_id="your_project_id",
    base_url="https://your-api-url.com",
)

prompt = prompt_manager.get_active_prompt_version(prompt_name="sampleprompt")
print(prompt)

Method parameters:

Parameter	Description
prompt_name	The name of the prompt as defined in Elsai Foundry

---

Using a prompt with an agent

Pass the retrieved prompt directly to an agent as its system prompt:

from elsai_prompts.prompt_manager import PromptManager
from elsai import Agent

prompt_manager = PromptManager(
    api_key="your_elsai_api_key",
    project_id="your_project_id",
    base_url="https://your-api-url.com",
)

prompt = prompt_manager.get_active_prompt_version(prompt_name="customer-support")

agent = Agent(system_prompt=prompt)
result = agent("How do I reset my password?")
print(result)

---

Version history

Version	Changes
1.1.0	Migrated from Pezzo to Elsai Foundry; PromptManager with project_id and base_url
1.0.0	Deprecated — Pezzo-based prompt management

Support

For SDK setup or API key access: support@elsaifoundry.ai