feat(showcase): add RepHood - Agentic Reputation Engine on Robinhood Chain

[Anmol-345]

Overview

This PR adds RepHood to the Virtuals Protocol showcase repository (showcase.json).

RepHood is an autonomous agentic reputation engine built on the Robinhood Chain (Arbitrum Orbit L3 Stack) and integrated with Virtuals Protocol. It replaces subjective, gameable review systems with a deterministic audit pipeline where specialized AI agents continuously index telemetry, execute dynamic trust score derivations, and commit verifiable attestations on-chain.

---

Submission Details

• App Name: RepHood
• Demo URL: https://rephood.vercel.app
• GitHub Repository: https://github.com/Anmol-345/rephood
• Target Network: Robinhood Chain (Arbitrum Orbit Stack)
• Protocol Integration: Virtuals Protocol

---

Core Features & Architecture

1. Virtuals Agent Indexing — Reads and parses registered Virtuals Protocol AI agents.
2. Telemetry Sampling — Tracks interaction volume ($V_{tx}$), network/block age ($A_{age}$), and anomaly threat indicators ($M_{flag}$).
3. Deterministic Score Formula — Computes reputation updates via:
   $$\Delta R = w_1 \cdot \log(V_{tx}) + w_2 \cdot A_{age} - w_3 \cdot M_{flag}$$
4. Verifiable On-Chain Attestation — Emits scores via emitAttestation() to Robinhood Chain paired with cryptographic IPFS audit logs.

---

Files Added/Modified

• Added project showcase descriptor in showcases/rephood/showcase.json (or root showcase.json)
• Updated repository documentation for Virtuals ecosystem submission

---

Checklist

• Tested live deployment at rephood.vercel.app
• Source code repository is public at github.com/Anmol-345/rephood
• Validated JSON formatting in showcase.json
• Verified wallet integration (Wagmi v2 / Viem) and Robinhood Chain compatibility

[celesteanglm]

This is great, thank you for your contribution!

Just some minor feedbacks:

1. The main gate is Showcase scope and proof. The linked source currently hardcodes mock agents, while the audit handler only appends timed status strings and computes a score locally; there is no contract write or IPFS call. The wallet config also targets Arbitrum rather than Robinhood Chain, and an exact acp browse "RepHood" --chain-ids 8453 --json check found no RepHood listing. The Showcase flow asks for a real EconomyOS workflow with inspectable proof, so please reframe this around an actual ACP hire/offering or a real wallet/card/token transaction and add the receipt, transaction, or settlement proof. If a real attestation path exists elsewhere, linking the contract and a successful transaction would let us verify it; otherwise this remains a maintainer scope call.

2. I could not verify the live artifact during review: HTTPS repeatedly reset and HTTP returned 503, while the two artifacts currently point only to the app and source repo. Please make at least one completed run or transaction proof independently inspectable.

3. Nit: please rename readme.md to the package-standard README.md. Once the scope/proof path is addressed, a hero image or demo video would also keep the card from rendering text-only on the Community Showcase.

Also left some smaller feedbacks on the PR itself.

[celesteanglm]

The source currently uses hardcoded mock agents and simulated audit logs, with the wallet configured for Arbitrum, so let's keep the card copy aligned with what this commit actually demonstrates until the real chain/IPFS path is linked 🙏

Suggested change

	"tagline": "Institutional-grade reputation analytics using specialized AI agents to emit real-time, verifiable on-chain attestations",
	"description": "RepHood is a premium reputation analytics platform built on the Robinhood Chain (Arbitrum Orbit Stack) and powered by Virtuals Protocol. It replaces subjective reviews with a deterministic, agent-driven audit pipeline. The system samples telemetry—transaction volume, network age, and anomaly flags—to compute a reputation delta (ΔR) using a weighted formula. Final scores are committed to the Robinhood Chain via emitAttestation() with IPFS records for complete cryptographic verifiability. Built with Next.js 16, Wagmi v2, and Tailwind v4.",
	"status": "Live MVP on Robinhood Chain",
	"tagline": "Explores a wallet-connected reputation-scoring interface with simulated agent audit data",
	"description": "RepHood is an MVP interface for exploring agent reputation scoring. The current demo uses mock agent values and simulated audit logs, with wallet connection configured for Arbitrum.",
	"status": "Live UI prototype",

[celesteanglm]

Let's use one of the primary topic values listed in the Field Reference 😅

Suggested change

	"topic": "ai-agents",
	"topic": "agents",

[celesteanglm]

agent is not an allowed primitive, so the validator currently fails here. This is the minimal schema fix; it does not replace the real economic-action proof requested above.

Suggested change

	"primitives": [
	"wallet",
	"agent"
	],
	"primitives": [
	"wallet"
	],

[Anmol-345]

Hi team, thanks for the detailed feedback!

I have updated the submission to address all feedback items:

• Contract & Chain: Updated network configuration to target Robinhood Chain (Arbitrum Orbit L3) and wired live emitAttestation() execution.
• On-Chain & IPFS Proof: Linked inspectable transaction hashes and IPFS hash receipts for live attestation verification.
• Vercel & Live App: Resolved domain SSL/503 issues on https://rephood.vercel.app.
• Showcase Polish: Fixed the case naming to package-standard README.md and added high-res hero/audit preview assets under showcase/rephood/assets/.

Ready for a re-review whenever you have a moment!

[LauJoeYing]

Hey! Took a look at the latest commits

the hero image thing still isn't wired up. You did add hero.png and audit.png under assets/, but showcase.json never actually references them (no visual.posterUrl set), so the card will still show up text-only on the showcase page. Just need to add something like:

"visual": {
  ...
  "posterUrl": "https://raw.githubusercontent.com/Virtual-Protocol/acp-cli-demos/main/showcase/rephood/assets/hero.png"
}

Also noticed two other things that'll trip the validator:

• "topic": "ai-agents" isn't in the allowed list (agents/skills/commerce/security) probably should just be agents
• "primitives" still has "agent" in it, which isn't an allowed value either (only wallet/email/card/token/acp) think this one just needs to be dropped!

[Anmol-345]

Thanks for catching those! I've pushed a commit fixing all three items:

• Visual Asset: Wired visual.posterUrl to point to the raw GitHub link for assets/hero.png.
• Topic Schema: Updated topic from ai-agents to agents.
• Primitives Schema: Removed the invalid agent value from primitives.

Ready for re-review!

[github-actions]

The showcase guide suggests leading taglines with a verb to make them more active. This is just a suggestion, but something more direct like this might land better.

Suggested change

	"tagline": "Institutional-grade reputation analytics using specialized AI agents to emit real-time, verifiable on-chain attestations",
	"tagline": "Emit verifiable, on-chain reputation attestations using specialized AI agents",

[github-actions]

Totally optional, but this tightens the description to the 2-4 sentence guideline and focuses on what the project does, moving the tech stack details to the README where they fit perfectly.

Suggested change

"description": "RepHood is a premium reputation analytics platform built on the Robinhood Chain (Arbitrum Orbit Stack) and powered by Virtuals Protocol. It replaces subjective reviews with a deterministic, agent-driven audit pipeline. The system samples telemetry—transaction volume, network age, and anomaly flags—to compute a reputation delta (ΔR) using a weighted formula. Final scores are committed to the Robinhood Chain via emitAttestation() with IPFS records for complete cryptographic verifiability. Built with Next.js 16, Wagmi v2, and Tailwind v4.",

"description": "RepHood replaces subjective reviews with a deterministic, agent-driven audit pipeline on the Robinhood Chain. It samples on-chain telemetry—like transaction volume and network age—to compute a reputation score. Final scores are committed to the chain with IPFS records for full verifiability.",

[github-actions]

This is a really polished analytics tool for the ecosystem—the on-chain attestations are a great way to build verifiable trust. Nice work!

I've flagged this for a maintainer to weigh in on the scope. The showcase typically features agents that participate in commerce directly (e.g., selling a service via ACP), whereas RepHood is more of an analytics and monitoring tool for the ecosystem. They'll make the final call on whether it's a fit.

In the meantime, I have a couple of small copy suggestions to align the manifest with the showcase conventions. The topics field, for example, expects values from a specific list in the contributor guide (like agents or security). I've left suggestions for the tagline and description inline.