personaai

Python SDK for the Persona.ai Developer Platform API — Agents, Skills, Knowledge bases, MCP connectors, and streaming chat, from your own backend.

Server-side only. Every method on this SDK sends your Project's credential — a server-side secret, not something client-side code is ever allowed to see. See Where do I call this from? for the full reasoning and a per-resource "who calls this, and when" table.

Install

pip install persona-agent-sdk

Installed as persona-agent-sdk on PyPI (the name personaai was already too close to an unrelated existing PyPI project); imported as personaai in code — from personaai import PersonaClient.

Requires Python 3.9+. Depends on httpx only — no requests, no aiohttp, no AG-UI protocol package (see Chat, streamed for why).

Quickstart

Ships both a sync client and an async client, built on the same httpx-based transport — use whichever matches your framework (Flask/Django vs. FastAPI/asyncio).

from personaai import PersonaClient

persona = PersonaClient(
    "https://api.persona.hasanraiyan.me",
    credential="<keyId>.<secret>",  # minted via Studio; a server-side secret, see warning above
)

# Sanity-check your credential.
who = persona.whoami()
print(who["principalType"], who["domain"])

# Provision an Agent (a one-time, control-plane call — no external user asserted).
agent = persona.agents.create(
    {
        "name": "Career Launchpad",
        "systemPrompt": "You help students find internships.",
        "providerId": "...",  # an existing Provider's id
        "visibility": "unlisted",
    }
)

Or async, identical shape:

import asyncio
from personaai import AsyncPersonaClient


async def main():
    async with AsyncPersonaClient(
        "https://api.persona.hasanraiyan.me", credential="<keyId>.<secret>"
    ) as persona:
        who = await persona.whoami()
        print(who["principalType"], who["domain"])


asyncio.run(main())

Both clients also work as context managers (with PersonaClient(...) as persona: / async with AsyncPersonaClient(...) as persona:), which closes the underlying httpx client for you — not required, but tidy for short-lived scripts.

Acting on behalf of one of your own end users

Most resources (Threads, Files, and any create/list call) behave differently depending on whether you assert an external user. Construct a second client per request, scoped to whoever is actually using your product right now — after your own auth has confirmed who that is:

user_persona = PersonaClient(
    "https://api.persona.hasanraiyan.me",
    credential="<keyId>.<secret>",
    external_user_id=current_user.id,  # your own user id for this person
)

thread = user_persona.threads.create({"agentId": agent["_id"]})

Chat, streamed

ChatClient.stream() is a regular generator; AsyncChatClient.stream() is an async generator — each is the natural per-language idiom for the same AG-UI event stream, keyword args (thread_id/resume) instead of one options object.

Events are typed as a loose dict (AguiEvent) rather than pulling in an AG-UI protocol package — every event has a "type" key (compare against the EventType string constants) plus type-specific fields (delta, name, value, ...). This mirrors the Node SDK's own call to reject @ag-ui/client's heavier dependency chain in favor of a hand-rolled SSE parser.

from personaai import EventType

# Full event stream, for building your own UI.
for event in user_persona.chat.stream(
    agent["_id"], [{"role": "user", "content": "What internships are open right now?"}]
):
    if event["type"] == EventType.TEXT_MESSAGE_CHUNK and event.get("delta"):
        print(event["delta"], end="")

# Or the convenience wrapper — drains the stream, returns the final text.
result = user_persona.chat.send_message(
    agent["_id"], [{"role": "user", "content": "What internships are open right now?"}]
)
print(result["text"])

# If the run pauses on a human-in-the-loop decision, `result["interrupt"]` is set instead of
# finishing normally — resume it on the next call:
if result["interrupt"]:
    user_persona.chat.send_message(
        agent["_id"],
        [],
        resume={"decisions": [{"action": "delete_agent", "decision": "approve"}]},
    )

Async is the same shape with async for/await:

async for event in user_persona.chat.stream(agent["_id"], messages):
    ...

result = await user_persona.chat.send_message(agent["_id"], messages)

Resources

Client property	Wraps
.agents	/api/v1/developer/agents
.skills	/api/v1/developer/skills
.knowledge	/api/v1/developer/knowledge (incl. document upload/search)
.mcps (+ .mcps.oauth)	/api/v1/developer/mcps (incl. OAuth owner/user connection flows)
.providers	/api/v1/developer/providers
.threads	/api/v1/developer/threads
.files	/api/v1/developer/files
.chat	/api/v1/developer/agui (streaming)

Every method mirrors the real REST endpoint 1:1 — no hidden behavior, and request/response bodies keep the API's own camelCase field names (providerId, systemPrompt, ...) even though method and parameter names are Pythonic snake_case. Types are TypedDicts exported from the package root — see each resource file under src/personaai/resources/ for the exact method signatures.

Out of scope for this SDK: Project/Members/Credentials management. Those are Clerk-session (human admin) operations, a completely different auth model than the machine-credential calls this SDK makes — manage them from Developer Studio instead.

Framework recipes

Connection pooling at high request volume

Constructing a PersonaClient/AsyncPersonaClient is cheap — no connection happens at construction time — so it's fine to build a fresh, per-request instance scoped to whoever is making the request (every recipe below does exactly that). But if you don't share anything, each instance opens its own httpx connection pool, so no TCP/TLS connection is ever reused across requests. At high volume, build one httpx.Client/httpx.AsyncClient at app startup and pass it to every per-request client via http_client= — they'll all share the same pool:

import httpx
from personaai import AsyncPersonaClient

# Built once, at app startup.
shared_http_client = httpx.AsyncClient()


def get_persona(external_user_id: str | None = None) -> AsyncPersonaClient:
    return AsyncPersonaClient(
        base_url,
        credential=credential,
        external_user_id=external_user_id,
        http_client=shared_http_client,  # reused across every call to get_persona()
    )

This is safe even if individual requests use with/async with on their own PersonaClient/AsyncPersonaClient — closing one of those never closes a client you passed in via http_client=; only a client the SDK created for you (no http_client given) gets closed. Close shared_http_client yourself, once, at app shutdown.

Flask

No special handling — construct the client once at module scope and use it in your view functions.

# persona.py
import os
from personaai import PersonaClient

persona = PersonaClient(os.environ["PERSONA_BASE_URL"], credential=os.environ["PERSONA_CREDENTIAL"])

# app.py
from flask import Flask, request, jsonify
from persona import persona
from personaai import PersonaClient
import os

app = Flask(__name__)


@app.post("/api/chat")
def chat():
    user_persona = PersonaClient(
        os.environ["PERSONA_BASE_URL"],
        credential=os.environ["PERSONA_CREDENTIAL"],
        external_user_id=request.json["userId"],
    )
    result = user_persona.chat.send_message(request.json["agentId"], request.json["messages"])
    return jsonify(result)

FastAPI

Construct the client per-request via a Depends() provider so it plugs into FastAPI's own dependency-injection system — the SDK itself needs no FastAPI-specific support, and this is the async client since FastAPI route handlers are async def. Shares one httpx.AsyncClient across every request (see Connection pooling at high request volume above) since a real API is likely to see meaningful request volume.

# deps.py
import os
import httpx
from personaai import AsyncPersonaClient

_shared_http_client = httpx.AsyncClient()


def get_persona(external_user_id: str | None = None) -> AsyncPersonaClient:
    return AsyncPersonaClient(
        os.environ["PERSONA_BASE_URL"],
        credential=os.environ["PERSONA_CREDENTIAL"],
        external_user_id=external_user_id,
        http_client=_shared_http_client,
    )

# main.py
from fastapi import Depends, FastAPI
from deps import get_persona
from personaai import AsyncPersonaClient

app = FastAPI()


@app.post("/api/chat")
async def chat(body: dict, current_user_id: str = Depends(get_current_user_id)):
    async with get_persona(external_user_id=current_user_id) as persona:
        return await persona.chat.send_message(body["agentId"], body["messages"])

Django

Wire a module-level singleton in apps.py (or a small persona.py module), same idea as Flask — Django views are sync by default, so the sync PersonaClient is the natural fit.

# yourapp/persona.py
from django.conf import settings
from personaai import PersonaClient

persona = PersonaClient(settings.PERSONA_BASE_URL, credential=settings.PERSONA_CREDENTIAL)

Keep PERSONA_CREDENTIAL in your environment / secrets manager and read it into settings.py (os.environ["PERSONA_CREDENTIAL"]) — never commit it, and never expose it via a TEMPLATE/context processor a browser can read.

Development

python -m venv .venv
.venv/Scripts/pip install -e ".[dev]"   # .venv/bin/pip on macOS/Linux
.venv/Scripts/pytest                     # unit tests (mocked via respx) — this is what CI runs
.venv/Scripts/mypy                       # mypy --strict
.venv/Scripts/ruff check .               # lint
.venv/Scripts/ruff format --check .      # format check

Integration tests (opt-in, needs a real backend)

tests/integration/test_live.py is skipped by default. It exercises the real SDK (both sync and async clients) against a real, already-running agent-backend — see the file's own header comment for the environment variables it needs (a real Project credential, and a real Provider id for the Agent/Knowledge/Threads tests). Every resource it creates is deleted by its own test; safe to run repeatedly against the same Project.

PERSONA_SDK_INTEGRATION_TEST=1 \
PERSONA_TEST_BASE_URL=https://api.persona.hasanraiyan.me \
PERSONA_TEST_CREDENTIAL=<keyId>.<secret> \
PERSONA_TEST_PROVIDER_ID=<provider-id> \
pytest tests/integration/test_live.py

Publishing

python -m build / twine upload (or flit publish) is not run as part of this repo's CI — releasing a new version to PyPI is a deliberate, separate action taken by a maintainer once a version is ready.

License

MIT