> For the complete documentation index, see [llms.txt](https://black-sheep-finance.gitbook.io/black-sheep-finance-docs/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://black-sheep-finance.gitbook.io/black-sheep-finance-docs/guides/multi-agent.md).

# Multi-agent budgets

Most real deployments use **several agent wallets**, each with its own scope and budget. This is intentional — it limits blast radius.

## Pattern: one agent per workflow

| Agent label     | Budget          | Allowlist                |
| --------------- | --------------- | ------------------------ |
| `research-bot`  | 50 USDC / day   | data vendors             |
| `support-bot`   | 200 USDC / day  | refund recipients (open) |
| `infra-bot`     | 1000 USDC / day | hosting providers        |
| `marketing-bot` | 500 USDC / day  | ad networks              |

If `research-bot` gets prompt-injected into trying to send 10,000 USDC to an attacker, the blast radius is 50 USDC / day, capped by allowlist.

## Pattern: per-customer agents

For multi-tenant SaaS where each customer's AI gets to spend:

* Issue one agent wallet per customer.
* Set `daily_limit_usdc` to their plan's spending cap.
* Use the API key per environment, not per customer.
* Use the `label` field as `customer:<id>` for easy filtering.

## Pattern: tiered approval

* **Tier 1 agent** — small amounts, fully autonomous (`approval_threshold_usdc` high).
* **Tier 2 agent** — medium amounts, on-call human approves (`approval_threshold_usdc` low).
* **Tier 3 wallet** — large amounts, **user wallet** (not agent), executive signs in dashboard.

## Don't

* Don't share one wallet across unrelated workflows. You lose attribution and one bad workflow can drain another's budget.
* Don't disable the approval threshold "for convenience". It's the one rule that catches the unknown unknowns.
* Don't fund an agent with more than \~1 week of its daily limit. Treat agent wallets like petty cash.


---

# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.

## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter, and the optional `goal` query parameter:

```
GET https://black-sheep-finance.gitbook.io/black-sheep-finance-docs/guides/multi-agent.md?ask=<question>&goal=<endgoal>
```

`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is optional and describes the broader end goal you are ultimately trying to accomplish on behalf of the user. GitBook uses it to tailor the answer towards what is most useful for that goal.

The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.