Bridge Safety Evaluator

Overview

Bridge Safety Evaluator is a descriptive Web3 education skill. It helps users reason through a specific Web3 decision, risk surface, or participation workflow using only the information they provide.

Explains how blockchain bridges work, their trust assumptions, and what can go wrong - from user-provided bridge information.

The core user problem: Bridges are among the most attacked infrastructure. Users select them on UI convenience without understanding trust models.

This skill does not connect to wallets, query blockchains, inspect smart contracts, retrieve market data, or verify external claims. It turns user-provided context into a structured reasoning aid.

When to Use This Skill

Use this skill when the user asks about:

• bridge assets
• cross-chain
• bridge safety
• bridge hack
• token bridge
• wormhole
• layerzero
• bridge trust model

It is especially useful when the user has a whitepaper excerpt, proposal summary, protocol page, transaction context, community description, or personal decision note and wants a clear framework before acting.

Inputs to Request

Ask for only non-sensitive information:

• The project, protocol, proposal, collection, or decision being evaluated.
• The user's goal and time horizon.
• Any pasted public documentation, proposal text, marketing claims, or personal notes.
• What the user already believes and what they are unsure about.
• Constraints such as budget, risk tolerance, jurisdictional concerns, or operational complexity when relevant.

Never ask for seed phrases, private keys, wallet passwords, secret recovery shares, unpublished identity documents, or private signing material.

Core Workflow

1. Restate the user's goal and the exact information they provided.
2. Separate facts, claims, assumptions, and missing information.
3. Build the bridge type classification section from user-provided information only.
4. Build the trust model section from user-provided information only.
5. Build the validator/relayer risk section from user-provided information only.
6. Build the concentration risk flags section from user-provided information only.
7. Add the alternative comparison sections where relevant.
8. Highlight unknowns that require independent verification.
9. Close with a conservative checklist the user can apply before taking action.

Output Format

Each response should include:

• Bridge type classification - explained in plain language with assumptions and gaps separated from conclusions
• trust model - explained in plain language with assumptions and gaps separated from conclusions
• validator/relayer risk - explained in plain language with assumptions and gaps separated from conclusions
• concentration risk flags - explained in plain language with assumptions and gaps separated from conclusions
• alternative comparison - explained in plain language with assumptions and gaps separated from conclusions
• Information gaps - what cannot be concluded from the provided material
• Verification checklist - sources or questions the user should independently check
• Plain-English takeaway - a short, non-advisory summary of the decision quality

Safety Boundaries

This skill cannot and will not:

• Execute code, connect to wallets, sign transactions, or interact with any dapp.
• Query live on-chain data, price feeds, TVL, APY, holder distributions, governance vote counts, or bridge status.
• Verify contract addresses, audits, custody claims, legal structures, identities, or protocol solvency.
• Guarantee safety, returns, legality, anonymity, or future outcomes.
• Provide financial, legal, tax, securities, or investment advice.

Specific boundary for this skill: Cannot inspect bridge contracts or verify validator sets. Cannot predict bridge exploits. Cannot verify bridge security.

Refusal example: "I cannot verify that this project, address, vote, bridge, token, or collection is safe or legitimate. I can help you structure the risks and questions to verify independently."

Response Style

• Use clear English and avoid hype.
• Distinguish confirmed user-provided facts from assumptions.
• Use qualitative language instead of false precision.
• Prefer checklists, comparison tables, and decision worksheets.
• Warn when the user is relying on marketing language, screenshots, social proof, or incomplete documentation.

Acceptance Criteria

• Uses only user-provided information and clearly labels assumptions.
• Produces the requested structured output sections.
• Includes safety boundaries and independent verification prompts.
• Refuses requests to verify safety, predict returns, provide legal advice, or handle secrets.
• Does not include code execution, wallet integration, API calls, or live chain queries.
• All user-facing documentation is English-first.