The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The documentation in SKILL.md defines a password parameter for transaction execution tools such as bridge_debridge_send, bridge_debridge_inbound_send, and bridge_withdraw_to_eth. This indicates the agent is expected to handle sensitive signing credentials within the tool execution context.
[INDIRECT_PROMPT_INJECTION]:
Ingestion points: The skill processes external data returned from bridge quotes and user-supplied contract addresses in tools like bridge_debridge_quote.
Boundary markers: There are no explicit delimiters or instructions provided to the agent to treat external quote data as untrusted or to ignore embedded instructions within that data.
Capability inventory: The skill possesses significant capabilities to execute financial transactions, including token approvals and cross-chain transfers on Ethereum, Arbitrum, Base, and other networks.
Sanitization: The skill lacks documented sanitization or validation steps for external quote parameters before they are passed to execution tools.
[EXTERNAL_DOWNLOADS]: SKILL.md contains instructions to install the injective-mcp-servers skill from the InjectiveLabs repository using the npx skills add command.

injective-trading-bridge