The Agent Skills Directory

[PROMPT_INJECTION]: The skill contains conflicting instructions between its core documentation and its implementation guides. Specifically, references/implementation.md directs the agent to use tool patterns like Bash(crypto:whale-*) which are not authorized in the SKILL.md frontmatter, and suggests reading credentials from an external .env file that is not part of the skill distribution.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection vulnerabilities.
Ingestion points: scripts/whale_api.py (Whale Alert and Etherscan APIs) and scripts/price_service.py (CoinGecko API) fetch data from external sources.
Boundary markers: None present. Data retrieved from external APIs is printed to the console without delimiters or instructions to the agent to disregard potential embedded commands.
Capability inventory: The agent is granted Bash execution privileges (restricted to python:*whale*), along with Read, Write, Edit, Grep, and Glob capabilities.
Sanitization: No sanitization or validation of API response content is performed before the data is presented to the agent.
[EXTERNAL_DOWNLOADS]: The Python scripts perform GET requests to api.whale-alert.io, api.etherscan.io, and api.coingecko.com. These are well-known services and the network activity is consistent with the skill's purpose.
[COMMAND_EXECUTION]: The skill's primary workflow involves the agent executing a Python CLI tool (scripts/whale_monitor.py) to query and analyze blockchain data.

monitoring-whale-activity