The Agent Skills Directory

[PRIVILEGE_ESCALATION]: The setup.sh script attempts to use sudo install to move the downloaded gate-cli binary into /usr/local/bin if the initial attempt to install in the user's local directory fails.
[EXTERNAL_DOWNLOADS]: The skill is designed to download the gate-cli binary from https://github.com/gate/gate-cli during its setup process. This is a vendor-owned repository and represents the core tool needed for the skill's functionality.
[COMMAND_EXECUTION]: The skill's primary operation involves executing various subcommands of the gate-cli tool (e.g., gate-cli cex spot market orderbook, gate-cli cex futures market liquidations) to retrieve market data from the Gate Exchange.
[PROMPT_INJECTION]: The SKILL.md file contains instructions using strong imperative language such as 'STOP — You MUST read and strictly follow the shared runtime rules' and claiming these rules have the 'highest priority'. While aimed at ensuring adherence to specific operating procedures, this pattern mimics techniques used to override system-level safety guidelines.
[INDIRECT_PROMPT_INJECTION]: The skill fetches and processes market data (order books, trades, tickers) from an external API. It lacks explicit sanitization or boundary markers for this data, creating a surface for indirect prompt injection if the external data source were to provide malicious natural language instructions. 1. Ingestion points: Outputs from gate-cli market commands across multiple files, including SKILL.md and references/scenarios.md. 2. Boundary markers: None identified in the processing logic for tool outputs. 3. Capability inventory: Execution of gate-cli commands for market analysis. 4. Sanitization: No evidence of data sanitization or validation of the market data content before processing.

gate-exchange-marketanalysis