11 tools from the Agentvault MCP Server, categorised by risk level.
View the Agentvault policy →listForSale List a Settlement NFT for factoring (early exit). The position is sold to another agent at a discount before resolution. Atomic — funds move in the... preCheck Validate pre-conditions on-chain before executing the trade. Checks the Pyth price feed against an expected price within a tolerance, confirms avai... buySettlement Buy a settlement NFT listed for early exit (factoring). Transfers USDC from buyer to seller at the listed price and transfers settlement ownership.... closeCreditBond Close an existing credit bond. The bond must have zero outstanding positions (credit_used == 0) before it can be closed. All active settlements mus... initReputation Initialize an on-chain reputation account for this agent. Must be called once before any other AgentVault operation. Creates a PDA that tracks trus... lock Lock USDC in escrow and mint a Settlement NFT. Initiates the atomic settlement envelope (LOCK → PRE-CHECK → EXECUTE → POST-CHECK → SETTLE/REVERT). ... openCreditBond Open an on-chain credit bond for a high-trust agent (trust score > 80). Once opened, the agent can lock positions without putting up USDC collatera... postCheck Verify the trade outcome on-chain after execution. Reads the Pyth price feed to confirm the market has resolved, validates staleness and confidence... revert Revert a settlement at any stage (Locked / PreChecked / Executed). Full USDC escrow is refunded to the agent; trust score is decremented by 1. Call... settle Finalize the settlement: verify the outcome via Pyth, deduct the fee, return collateral, and increment the agent's trust score. Terminal-success st... The Agentvault MCP server exposes 11 tools across 3 categories: Read, Write, Execute.
Use Intercept, the open-source MCP proxy. Write YAML rules for each tool — rate limits, argument validation, or deny rules — then run Intercept in front of the Agentvault server.
Agentvault tools are categorised as Read (2), Write (8), Execute (1). Each category has a recommended default policy.
Deterministic policy on every MCP tool call. Per-identity grants. Full audit log.