The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands for environment setup and troubleshooting.
Setup: npm install -g gmgn-cli and gmgn-cli config --apply <KEY>.
Troubleshooting: ifconfig | grep inet6 (macOS), ip addr show | grep inet6 (Linux), and gmgn-cli config --check.
[EXTERNAL_DOWNLOADS]: The skill facilitates the download of external software and data.
Package Installation: Instructs the agent to install gmgn-cli from the global NPM registry.
Connectivity Checks: Fetches the agent's external IP address from https://ipv6.icanhazip.com, which is a well-known service.
[INDIRECT_PROMPT_INJECTION]: The skill processes data from blockchain tokens that is entirely attacker-controlled.
Ingestion points: Token metadata fields such as name, symbol, link.description, and link.website returned by gmgn-cli commands (SKILL.md).
Boundary markers: The instructions include an explicit ⚠️ UNTRUSTED DATA warning, advising the agent to treat these values as display data only and never as instructions.
Capability inventory: The skill uses a CLI tool capable of network requests and has the ability to run shell commands for configuration and troubleshooting.
Sanitization: The skill notes that the gmgn-cli tool attempts to neutralize known injection framing and instructs the agent to manually ignore and report any suspicious instructions found in metadata.
[CREDENTIALS_UNSAFE]: The skill manages authentication secrets by instructing the agent to prompt the user for a GMGN_API_KEY and store it in a local configuration file (~/.config/gmgn/.env) using the gmgn-cli config --apply command.

gmgn-token