The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill requires fetching code and packages from external, non-whitelisted sources during installation.
Evidence: Installation instructions for the @emblemvault/agentwallet npm package.
Evidence: Skill acquisition via npx skills add EmblemCompany/Agent-skills.
[COMMAND_EXECUTION]: The setup process involves executing shell commands to install external dependencies.
Evidence: npm install @emblemvault/agentwallet and npx skills add commands specified in the setup section.
[DATA_EXFILTRATION]: The skill performs network operations to an external API service not included in the standard whitelist.
Evidence: Connection to https://api.agenthustle.ai for balance checks, swaps, and transfers.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection where malicious instructions could be embedded in blockchain metadata.
Ingestion points: Data retrieved from GET /portfolio/{address} and GET /token/{chain}/{contract} (SKILL.md).
Boundary markers: The skill includes a requirement to 'Always confirm before executing transactions', providing a human-in-the-loop safety check (SKILL.md).
Capability inventory: The agent can perform high-impact actions like POST /swap and POST /transfer (SKILL.md).
Sanitization: There is no explicit evidence of sanitization for token names or other external metadata processed by the agent.

emblemai-crypto-wallet