Axme

I've been building AXME for the past year - an intent-based protocol and managed service for operations that finish later.

The problem: every time I needed a human to approve something mid-workflow, I'd build the same glue - webhook endpoint, email sender, polling loop, Redis for state, retry logic. 200 lines before the actual business logic.

With AXME it's 4 lines:
intent_id = client.send_intent(to="agent://my-corp/service", ...) result = client.wait_for(intent_id)

AXME handles routing, delivery, retries, timeouts, reminders, and human approval gates.

What I built:
5 SDKs (Python, TypeScript, Go, Java, .NET), CLI with 30+ built-in runnable examples, MCP server with 28 tools for AI assistants, 8 human task types (approval, form, review, override, confirmation, assignment, clarification, manual action), 5 delivery modes (SSE stream, poll, HTTP webhook, inbox, internal runtime).

Runs on GCP - 5 Cloud Run services, Pub/Sub, PostgreSQL, BigQuery. No Redis, no Kafka. Solo founder, built everything from protocol spec to deployment.

Not a Temporal replacement - different model. No determinism constraints, no workflow engine to operate. Simpler for the 80% of cases where you need "do thing, maybe ask human, finish later."

Alpha stage. Looking for feedback and early adopters.
GitHub: github.com/AxmeAI/axme
Quick start: cloud.axme.ai/alpha/cli