Spuro x HTTPayer · Sub0 Hackathon

Milestone 2 Plan: Spuro – Arkiv-backed Agent SDK

This page follows the Sub0 Milestone 2 template, adapted to the Spuro docs & catalogue demo.

sub0 Hackathon Demo

This proof-of-concept demonstrates agent-native infrastructure: AI agents use Arkiv-backed memory via Spuro and HTTPayer (x402), while an external agent (e.g. AI-Spuro) queries Polkadot via PAPI and persists stash-account snapshots into that paid memory layer.

Arkiv Memory BucketHTTPayer x402PAPI (external)

📍 Where we are now

What we built/validated this weekend

  • Spuro backend: FastAPI service wrapping Arkiv with x402/HTTPayer, exposing paid entity endpoints.
  • Spuro Functions SDK in packages/ts, providing typed helpers like createEntity and queryEntities.
  • AI-Spuro demo agent that snapshots Polkadot stash accounts via PAPI and persists them to Arkiv through Spuro.

What's working

  • End-to-end flow from external data source → AI-Spuro agent → Spuro → Arkiv, with x402 handling payments at the HTTP layer.
  • Docs & catalogue UI that explains the core Spuro endpoints and the Spuro Functions used by AI-Spuro.

What still needs work

  • Turning AI-Spuro into a reusable Spuro Agent SDK so other teams can plug in their own prompts and flows.
  • Sharper agent patterns for Arkiv-backed memory (TTL, session-scoped buckets, query helpers).

Blockers or hurdles we hit

  • Balancing time between core infra (Spuro/Arkiv/x402) and higher-level DX (SDK ergonomics, agent templates).
  • Designing abstractions that work both for Polkadot data and non‑Polkadot data sources without over-complicating the SDK.

🚀 What we'll ship in 30 days

Our MVP will do this

Ship an AI SDK for Spuro that any TypeScript agent can import to get Arkiv-backed memory, x402-paid storage, and ready-made helpers for snapshotting external data (starting with Polkadot accounts). For hackathon teams and future users, this means they can wire "agent memory as a service" into their bots in under an hour.

Features we'll build (3–5 max)

Weeks 1–2

Feature: Spuro Agent SDK package

Why it matters: turns today's Spuro Functions into a consumable SDK with clear typings, examples, and sensible defaults for agents.

Who builds it: backend/SDK engineer (Spuro + Arkiv).

Weeks 2–3

Feature: Agent memory patterns & helpers

Why it matters: gives teams opinionated recipes for snapshots, TTL, and querying so they don't have to design their own Arkiv schema.

Who builds it: backend + agent-experience engineer.

Weeks 3–4

Feature: Example agents & docs

Why it matters: shows how to plug the SDK into concrete agents (Polkadot stash monitor plus a generic snapshotting bot) and how to run them end-to-end.

Who builds it: frontend/docs + agent-experience engineer.

Team breakdown

Backend Engineer – Spuro/Arkiv | ~10 hrs/week

Owns: Spuro API surface, Arkiv entity design, SDK core.

Infra/Payments Engineer – HTTPayer/x402 | ~8 hrs/week

Owns: x402 payment flows, API security, example configs for paid agents.

Agent Experience Engineer – SDK DX & Docs | ~8 hrs/week

Owns: AI-Spuro CLI demos, docs site, agent templates, and onboarding guides.

Mentoring & expertise we need

Areas where we need support

  • Validating our SDK design against how other agent frameworks structure tools and memory.
  • Stress-testing Arkiv/Spuro patterns for scale and long-lived agent sessions.

Specific expertise we're looking for

  • Mentors with experience building agent SDKs or tool ecosystems.
  • Experts in Arkiv / on-chain data storage best practices to review our entity and TTL design.

🎯 What happens after

When M2 is done, we plan to...

  • Open up the Spuro Agent SDK to early design partners (other hackathon teams, infra projects) and iterate on feedback.
  • Add 1–2 additional example agents beyond Polkadot, showing how Spuro-backed memory can power any autonomous workflow.

And 6 months out we see our project achieve:

  • Multiple external teams shipping production agents that rely on Spuro/Arkiv as their primary memory layer.
  • Spuro recognized as a core building block for agent-native infrastructure, with HTTPayer/x402 and Arkiv underneath and a healthy ecosystem of community-contributed agent templates on top.