The Agent Skills Directory

[COMMAND_EXECUTION]: The install.sh and install_cli.sh scripts utilize shell commands to detect AI platform environments, configure local directories (e.g., ~/.cursor/skills), and modify configuration files using jq and sed.
[EXTERNAL_DOWNLOADS]: The install_cli.sh script performs a global installation of the gate-wallet-cli package via npm. This is a vendor-provided tool required for command-line functionality.
[PROMPT_INJECTION]: The SKILL.md file contains strong imperative language and priority-override instructions, such as "⚠️ STOP — You MUST read and strictly follow..." and "These rules have the highest priority." These are safety-oriented instructions meant to direct the agent to specific runtime guidelines.
[INDIRECT_PROMPT_INJECTION]: The skill exposes an attack surface by processing untrusted data from external sources, including DApp contract data and arbitrary HTTP 402 payment URLs.
Ingestion points: Data ingested via dex_tx_x402_fetch (URL, headers, body) and DApp contract ABIs/transaction data.
Boundary markers: Explicit user confirmation gates are mandatory for all signing operations, requiring the agent to wait for a "confirm" response from the user.
Capability inventory: The skill possesses transaction signing (dex_wallet_sign_transaction), network broadcasting (dex_tx_send_raw_transaction), and arbitrary URL fetching (dex_tx_x402_fetch) capabilities.
Sanitization: The skill implements a mandatory "Gate Verify MCP" check-in (tx_checkin) which acts as a secondary verification layer before any signing call.

gate-dex-wallet