Pythathon - Guards | Risk Leveled Treasury

[f0x1777]

Pyth Examples Contribution

Team Name: Guards
Submission Name: Guards One
Team Members: @f0x1777 @kevan1 @joaco05
Contact: @f0x1777

Type of Contribution

• New Example Project (Adding a new example to demonstrate Pyth integration)
• Bug Fix (Fixing an issue in existing examples)
• Documentation Update (Improving README, comments, or guides)
• Enhancement (Improving existing functionality or adding features)
• Hackathon Submission (Submitting a project from a hackathon)

Project Information

Project/Example Name: Guards One

Pyth Product Used:

• Pyth Price Feeds
• Pyth Entropy
• Multiple Products
• Other: ___________

Blockchain/Platform:

• Cardano
• Ethereum/EVM
• Solana
• Aptos
• Sui
• Fuel
• Starknet
• TON
• Other: ___________

Description

What does this contribution do?

Guards is a multichain treasury protection workflow focused on monitoring liquid portfolio value, fiat/stable floors, and automated de-risking conditions. This submission focuses on the Cardano deployment surface.

How does it integrate with Pyth?

It uses Pyth price feeds as the oracle layer for treasury risk evaluation and oracle-aware execution decisions, with Cardano as the primary live surface for this hackathon submission.

What problem does it solve or demonstrate?

It demonstrates how a treasury can use Pyth data to protect stable-denominated value and react to market deterioration with objective rules across chains, while delivering the Cardano implementation in this submission.

Testing & Verification

How to Test This Contribution

The current implementation can be verified in three layers:

1. Static checks: typecheck and unit/integration tests
2. Local simulation: deterministic breach -> de-risk -> exit -> recovery flow
3. Live oracle wiring: fetch a real signed Pyth update for Cardano preprod

Prerequisites

• Node.js >= 24.0.0
• pnpm
• A .env file based on .env.example
• A valid PYTH_API_KEY
• Cardano preprod configuration for live verification:
  • PYTH_PREPROD_POLICY_ID
  • CARDANO_BLOCKFROST_PROJECT_ID
  • CARDANO_PYTH_STATE_REFERENCE

Setup & Run Instructions

pnpm install
cp .env.example .env

# Static validation
pnpm typecheck
pnpm test

# Deterministic treasury simulation
pnpm simulate

# Fetch a live signed Pyth update for the Cardano preprod flow
pnpm pyth:fetch-live

# Run the UI locally
pnpm --dir apps/ui dev

Local UI routes:

• Landing: http://localhost:3000
• Dashboard: http://localhost:3000/dashboard

Deployment Information (if applicable)

Current deployment target for the hackathon is Cardano preprod.

Relevant deployment/runtime notes:

• Pyth preprod policy id:
  d799d287105dea9377cdf9ea8502a83d2b9eb2d2050a8aea800a21e6
• Frontend can be deployed from apps/ui
• Off-chain oracle/keeper services require a Node 24 runtime plus the Cardano/Pyth environment variables listed above
• The multichain control-plane logic is shared, but this submission's live execution path is centered on Cardano preprod

Checklist

Code Quality

• Code follows existing patterns in the repository
• Proper error handling implemented
• No hardcoded values (use environment variables where appropriate)

Testing

• Tested locally and works as expected
• All existing functionality still works (no breaking changes)

Additional Context

Related Issues

N/A

Screenshots/Demo (if applicable)

Not added yet.

Notes for Reviewers

Draft bootstrap only. This PR currently contains the initial project description under lazer/cardano/guards/ and will be updated incrementally.

---

[f0x1777]

Please dismiss last two commits, was uploaded by the agent after submission time when was connected again to internet

[devin-ai-integration]

Devin Review found 1 potential issue.

View 5 additional findings in Devin Review.

[devin-ai-integration]

🟡 Reference price input disabled by stale/wide quotes that fail oracle guardrails

The currentReferencePrice prop passed to VaultBootstrapLab is computed WITHOUT oracle guardrails (dashboard/page.tsx:345-348), while the useEffect that updates bootstrapDraft.referencePrice applies guardrails (dashboard/page.tsx:134-140). When a live quote exists but becomes stale or exceeds confidence thresholds, currentReferencePrice still returns a number (the stale price), causing liveReferenceAvailable to be true at vault-bootstrap-lab.tsx:71-72. This disables manual editing of the reference price field (vault-bootstrap-lab.tsx:365), but the useEffect refuses to sync the stale value into the draft. The result: the input displays a stale value and is locked, preventing the user from overriding it, while the draft holds a different (older but guardrail-passing) value that's actually used by the scenario lab.

Root cause in dashboard/page.tsx call site

The unguarded call at line 345-348:

currentReferencePrice={liveReferencePriceForSymbol(
    liveQuotes ?? {},
    bootstrapDraft.referenceSymbol,
)}

should match the guarded call at line 134-140 by passing { maxStaleUs: demoState.policy.maxStaleUs, maxConfidenceBps: demoState.policy.maxConfidenceBps } as the third argument.

Suggested change

	currentReferencePrice={liveReferencePriceForSymbol(
	liveQuotes ?? {},
	bootstrapDraft.referenceSymbol,
	)}
	currentReferencePrice={liveReferencePriceForSymbol(
	liveQuotes ?? {},
	bootstrapDraft.referenceSymbol,
	{
	maxStaleUs: demoState.policy.maxStaleUs,
	maxConfidenceBps: demoState.policy.maxConfidenceBps,
	},
	)}

---

Was this helpful? React with 👍 or 👎 to provide feedback.