Check an agent's Valta wallet balance and spend limits. Call this before attempting a spend to confirm funds are available. This does not spend anything.

Request authorization to spend from an agent's Valta wallet. Returns approved with a new balance, or denied with a reason. IMPORTANT: this tool only governs spend that routes through it — it does not intercept or prevent spending done via other tools, APIs, or webhooks that don't call this tool first. If denied, do not proceed with the action and inform the user why.

Create a new named Valta wallet with optional spend limits. Use this to set up governance for a new agent before it starts spending.

Immediately freeze an agent's Valta wallet, blocking all further spend through Valta. Use this as a kill switch when an agent is behaving unexpectedly.

Unfreeze a previously frozen agent's Valta wallet, allowing spend to resume.

Read the tamper-evident, hash-chained audit trail of spend decisions (approved and denied) for an agent, or across all agents if no agentId is given.

List all named Valta wallets on this account, with balances and limits. Use this to discover what wallets exist before checking a balance or requesting a spend — you need to know the exact wallet name/ID first.

Transfer USDC directly between two of your agent wallets.

List all agents on this account — both marketplace subscriptions and custom SDK-created agents.

Get details for a single agent, including status and balance.

Trigger an agent to run a task. Returns the run's execution ID and initial status.

Check the status and result of a specific agent run.

List spending policies (limits, approval thresholds) configured on this account, optionally filtered to one agent.

Create a new spending policy — a named limit configuration (daily cap, per-transaction cap) that can be applied to an agent's spend.

Make a real outbound API call to a connected service (e.g. Stripe) through Valta's spend-gated egress proxy, instead of calling the provider directly. Declare the amount this call will cost up front — Valta checks it against the agent's wallet limits (freeze state, per-transaction/daily/monthly caps, approval thresholds) and only makes the real request if approved, using the account's own stored credential for that service. If the upstream call fails, the declared amount is automatically refunded. Only services with proxying enabled can be called this way — currently: stripe, serper, polygon.

List the receipts for every call made through Valta's spend-gated egress proxy — what was allowed, blocked, refunded, or left pending approval, and why. Optionally filter to one agent.