RLUSD Skills
RLUSD Skills route agent prompts into safe, structured RLUSD workflows.
RLUSD Skills is the agent-facing skill and documentation layer for RLUSD workflows. It packages prompt routing, workflow instructions, reference material, and tests that keep agent behavior aligned with the external RLUSD CLI runtime.
The repository is intentionally skills-and-docs focused. Runtime execution belongs to RLUSD CLI. This separation lets the skill layer teach agents how to route a request while keeping chain metadata, wallet execution, and command behavior in a deterministic tool.
This design reflects T54's broader view of agentic finance. Agents should not improvise financial operations from natural language alone. They should operate through explicit skills, structured plans, and runtime tools that can be reviewed and connected to risk controls.
Current Coverage
RLUSD Skills covers several workflow families.
| Skill area | Examples |
|---|---|
| Wallet preflight | Inspect or set up local wallet aliases before wallet-backed actions. |
| XRPL | Trust-line status, trust-line prepare and execute, account info, payment prepare and execute. |
| Ethereum | RLUSD metadata, transfer and approval preparation, transaction monitoring. |
| DeFi | Venue discovery, live swap quotes, prepared swaps, Curve LP workflows, and Aave supply flows. |
| Fiat guidance | Onboarding, buy, and redeem guidance. |
| x402 | Fetch x402-protected resources with an XRPL wallet and a spending cap. |
The skill layer encourages agents to follow the same operational discipline as human operators: inspect the context, prepare the action, review the plan, execute only when required, and verify the result. That discipline is what makes RLUSD workflows compatible with future Trustline-governed risk and underwriting paths.
Safety Model
Most write workflows follow prepare -> review -> execute. Prepared plans are stored by the runtime and validated before execution. This pattern is important for agentic finance because it prevents a prompt from becoming an immediate irreversible transaction without a reviewable artifact.
The x402 fetch flow is different. It is a capped direct-execution payment negotiation flow. The agent must provide a positive --max-value, and the CLI selects compatible x402 payment requirements under that cap.
Example x402 Flow
rlusd config get --json rlusd wallet list --json rlusd x402 fetch https://api.example.com/premium-feed \ --wallet ops-xrpl \ --max-value 1 \ --json
For RLUSD-only flows, the agent can constrain the advertised payment requirement by asset and issuer.
rlusd x402 fetch https://api.example.com/premium-feed \ --wallet ops-xrpl \ --max-value 1 \ --require-asset RLUSD \ --require-issuer rBvKgF3jSZWdJcwSsmoJspoXGpHUhBGurg \ --json
The workflow accounts for the fact that XRPL servers may advertise RLUSD using the canonical 40-hex currency code rather than the human-readable symbol. Agents inspect the accepts array when negotiation fails.
Why It Matters
RLUSD Skills make RLUSD workflows understandable to agents without letting agents improvise chain mechanics. The result is a safer bridge between natural-language requests and deterministic RLUSD operations.