The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill downloads an installation script and its core binary from the official vendor repository on GitHub. Integrity is maintained through SHA256 checksum verification against a signed checksum file before execution.- [COMMAND_EXECUTION]: The skill invokes the onchainos CLI and shell utilities to manage its lifecycle and perform blockchain transactions. It implements a mandatory confirmation gate for any command that executes a write operation on-chain.- [PROMPT_INJECTION]: The skill processes data from external DeFi APIs.
Ingestion points: Data enters via src/api.rs (Morpho GraphQL) and src/commands/claim_rewards.rs (Merkl API).
Boundary markers: The SKILL.md defines a 'Data Trust Boundary' explicitly instructing the agent to treat CLI output as untrusted external content.
Capability inventory: The skill can broadcast on-chain transactions via onchainos wallet contract-call.
Sanitization: The binary filters API responses to human-relevant fields, and the SKILL.md directs the agent to only display specific fields (asset, amount, APY) to the user.

morpho-plugin