The Agent Skills Directory

[SAFE]: The skill provides documentation and implementation examples for a cross-chain swap service.
[EXTERNAL_DOWNLOADS]: The code examples reference well-known, trusted libraries from standard package registries (NPM) such as viem, wagmi, and various blockchain SDKs (Solana, Stellar, Ton, etc.).
[COMMAND_EXECUTION]: The server-example.md file contains a command to run the script using npx tsx, which is standard for Node.js development.
[CREDENTIALS_UNSAFE]: The skill correctly instructs users to manage sensitive data like private keys and API keys through environment variables (process.env), which is an industry-standard safe practice. No hardcoded credentials were found.
[DATA_EXFILTRATION]: Network operations are directed towards the official service domain (1click.chaindefuser.com) and well-known blockchain RPC endpoints (NEAR, Solana, Stellar, etc.). No unauthorized data transmission or exfiltration patterns were observed.
[PROMPT_INJECTION]: No attempts to override agent behavior or bypass safety guidelines were found in the instructions or metadata.
[OBFUSCATION]: No hidden content, encoding tricks, or homoglyph attacks were detected.
[INDIRECT_PROMPT_INJECTION]: The skill processes data from external API endpoints (token lists, swap quotes). While this presents a theoretical attack surface for indirect prompt injection, the risk is inherent to the functional requirements of the service, and the skill follows standard implementation patterns.
Ingestion points: rules/api-tokens.md (GET /v0/tokens), rules/api-quote.md (POST /v0/quote)
Boundary markers: Absent
Capability inventory: Subprocess calls (via tsx in rules/server-example.md), network operations (fetch), and blockchain transactions (multiple files)
Sanitization: Standard API client logic with no explicit sanitization shown in the snippets

near-intents