The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill interacts with several well-known blockchain and bridge APIs to retrieve chain metadata, transaction history, and bridge statuses.
Evidence: Queries are made to official domains including etherscan.io, blockscout.com, bungee.exchange, li.quest, and layerzero-api.com.
Context: These services are industry-standard infrastructure providers for the EVM ecosystem.
[COMMAND_EXECUTION]: The skill utilizes local shell scripts and CLI tools to perform plan detection and chain resolution.
Evidence: Scripts such as scripts/etherscan-detect-plan.sh, scripts/blockscout-detect-plan.sh, and scripts/resolve-chain.sh use curl to probe API limits and metadata.
Context: These are utility scripts designed to optimize API usage and ensure correct chain mapping, representing standard integration practices.
[PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection by processing external data from blockchain explorers and bridge APIs.
Ingestion points: JSON responses from external APIs (e.g., Etherscan, Blockscout) are parsed into the agent's context.
Boundary markers: The instructions explicitly command the agent to treat these sources as read-only enrichment and strictly forbid signing messages or broadcasting transactions based on the retrieved data.
Capability inventory: Capabilities include shell execution via curl and cast, and reading local reference files.
Sanitization: The skill includes logic for unit conversion (wei to human-readable) and strict chain-id validation against an authoritative local list.

evm-atlas