explaingit

exakoss/shadowpilot

Analysis updated 2026-05-18

0TypeScriptAudience · developerComplexity · 5/5Setup · hard

TLDR

A hackathon project where verified humans get paid in Solana to remotely rescue stuck robots, with each intervention captured as licensable AI training data.

Mindmap

mindmap
  root((ShadowPilot))
    What it does
      Human robot recovery
      Pays pilots in Solana
      Captures training traces
    Tech stack
      TypeScript
      Next.js
      Solana
      Anchor
    Use cases
      Teleoperation marketplace
      Robot fallback tasks
      Physical AI datasets
    Audience
      Robotics teams
      Web3 developers
    Protocol
      World ID
      Arcium
      Metaplex NFTs

Code map

Detail Auto

An interactive map of this repo's files and how they connect — its source is parsed live in your browser. Click Visualize to build it.

filefunction / class

What do people build with it?

USE CASE 1

Post a funded fallback task for a robot that got stuck and have a verified human resolve it remotely.

USE CASE 2

Earn payment in Solana by acting as a pilot who completes teleoperation recovery tasks.

USE CASE 3

Collect private, licensable human intervention data to train physical AI or robotics models.

USE CASE 4

Prototype a private data provenance and payout system for robotics teleoperation using Solana.

What is it built with?

TypeScriptNext.jsSolanaAnchorRust

How does it compare?

exakoss/shadowpilot0xradioac7iv/tempfsabboskhonov/hermium
Stars000
LanguageTypeScriptTypeScriptTypeScript
Setup difficultyhardmoderatemoderate
Complexity5/53/54/5
Audiencedeveloperdeveloperdeveloper

Figures from each repo's GitHub metadata at analysis time.

How do you get it running?

Difficulty · hard Time to first run · 1h+

Requires Node.js 20+, pnpm 10+, Rust, Cargo, and Solana tooling to run the full stack.

In plain English

ShadowPilot is a TypeScript platform that creates a marketplace for human-assisted robot recovery. When an autonomous robot gets stuck in a situation it cannot handle, ShadowPilot lets a verified human operator step in remotely, complete the recovery task, and receive payment in Solana. The human's intervention is preserved as a signed, rights-bearing training data trace rather than disappearing into an operations log. The platform connects three workflows: a buyer (a robotics team) creates a task and deposits funds in an on-chain escrow, a pilot (a human operator) claims the task, performs the teleoperation to resolve the situation, and submits a trace, then confidential scoring evaluates the quality of the intervention privately before releasing payment and issuing a training-rights receipt to the buyer. That receipt is a compressed NFT representing the buyer's right to use the intervention data for training AI models. Protocol components include Solana and the Anchor smart contract framework for task accounts, escrow, and payout, World ID for optional proof that pilots are unique humans, Arcium for private quality scoring without exposing raw task data, and Metaplex compressed NFTs for training-rights receipts. The current MVP focuses on a simulated warehouse rover scenario. The web application provides separate interfaces for buyers, pilots, and a mission overview, built with Next.js. The project was built for the Solana Colosseum hackathon and is described in the README as not production software.

Copy-paste prompts

Prompt 1
Help me set up ShadowPilot locally with pnpm and run through the buyer and pilot flows.
Prompt 2
Explain how ShadowPilot uses Arcium for private scoring without exposing raw task data.
Prompt 3
Walk me through ShadowPilot's Anchor program for task accounts, escrow, and payout.
Prompt 4
Show me how the training-rights receipt is minted as a compressed NFT after an accepted intervention.
Prompt 5
Help me understand ShadowPilot's core flow from task funding to pilot payout.

Frequently asked questions

What is shadowpilot?

A hackathon project where verified humans get paid in Solana to remotely rescue stuck robots, with each intervention captured as licensable AI training data.

What language is shadowpilot written in?

Mainly TypeScript. The stack also includes TypeScript, Next.js, Solana.

How hard is shadowpilot to set up?

Setup difficulty is rated hard, with roughly 1h+ to a first successful run.

Who is shadowpilot for?

Mainly developer.

Open on GitHub → Explain another repo

This repo across BitVibe Labs

Verify against the repo before relying on details.