The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill utilizes the @aomi-labs/client package from the NPM registry to perform transaction construction and network interactions.
[COMMAND_EXECUTION]: Leverages the aomi command-line interface to execute blockchain-related tasks including session management, transaction listing, and broadcasting.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted natural language input to generate on-chain transactions, which represents a potential injection surface.
Ingestion points: Natural language prompts passed to the aomi chat command as documented in SKILL.md.
Boundary markers: Instructions in SKILL.md explicitly forbid the agent from including signing commands in automated blocks and require a separate confirmation step.
Capability inventory: The skill can simulate, sign, and broadcast transactions through the Aomi CLI as utilized throughout SKILL.md.
Sanitization: SKILL.md describes internal security features that block transactions where recipients do not match the sender to prevent fund drainage.
[CREDENTIALS_UNSAFE]: Documentation in SKILL.md describes the aomi secret add utility for managing application-specific credentials, which is a standard mechanism for secure secret storage within the tool's environment.

aomi-transact