pushary-crewai

Full walkthrough: Human-in-the-loop for CrewAI. Reaching your own end-users on their phones is the Pushary Partner plan.

Human-in-the-loop for CrewAI. Replace the console human_input=True prompt with a tool that reaches a real person on their phone and blocks until they answer, fail-closed.

Requires the Pushary Partner plan.

Install

pip install pushary-crewai

Set PUSHARY_API_KEY (get it in your dashboard).

Connect a phone once

from pushary_crewai import connect

link = connect("user_123")  # show this to your end-user; one tap connects their phone

Give an agent an ask-human tool

from crewai import Agent, Task, Crew
from pushary_crewai import make_ask_human_tool

agent = Agent(
    role="Ops",
    goal="Ship safely",
    backstory="Careful operator.",
    tools=[make_ask_human_tool("user_123")],
)
task = Task(
    description="Draft the release. Before finalizing, call ask_human to get approval.",
    expected_output="approved release notes",
    agent=agent,
)
Crew(agents=[agent], tasks=[task]).kickoff()

The tool blocks until the person answers and returns a fail-closed instruction. The external_id is bound when you build the tool, never taken from the model, so a prompt-injected agent cannot ask the wrong person.

Lower-level helpers

from pushary_crewai import ask_human

d = ask_human("Approve this refund?", external_id="user_123", type="confirm")
if d["approved"]:
    issue_refund()

API

• connect(external_id, *, api_key=None, base_url=None) — enroll an end-user's phone.
• make_ask_human_tool(external_id, *, name=..., ...) — a CrewAI BaseTool bound to that user.
• ask_human(question, *, external_id, type="confirm", ...) — blocking, returns the decision dict.
• resolve_pushary_callback(raw_body, signature, secret) — verify + parse a callback for a durable path.
• describe_answer(type, result), is_affirmative(answer), deterministic_key(parts), SIGNATURE_HEADER.

Example

A runnable example is in examples/.

License

MIT