> ## Documentation Index
> Fetch the complete documentation index at: https://redbark.com/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# Workflows, prompts, and guides

> Use composite tools and guided prompts to complete common Redbark tasks in fewer calls.

For multi-step tasks, your AI can use composite tools instead of assembling every API call itself. Prompts supply a task brief; guides explain the sequence and prerequisites.

## Composite tools

| Tool               | What it does                                                              | When to use it                             |
| ------------------ | ------------------------------------------------------------------------- | ------------------------------------------ |
| `account_overview` | Groups accounts by institution and fetches banking balances               | "What do my accounts look like right now?" |
| `spending_summary` | Totals a month of spending by category and account                        | "What did I spend in July?"                |
| `setup_sync`       | Creates a sync against an active destination and queues the first run     | After choosing a destination and accounts  |
| `diagnose_sync`    | Reads a sync, recent runs, connection, and destination to recommend a fix | "Why is my Sheets sync failing?"           |
| `connect_bank`     | Finds an institution and creates a link session                           | "Connect my ANZ account"                   |

Bank and destination authorisation still require you to complete the returned link. The AI cannot finish the consent screen for you. Use [setup](/docs/mcp/setup) to connect your AI client first.

## Prompts

Clients that support MCP prompts can offer these as slash-style commands. They insert a brief telling the AI which tools to use and how to report the result.

| Prompt                    | Arguments                                      | What it covers                                                                              |
| ------------------------- | ---------------------------------------------- | ------------------------------------------------------------------------------------------- |
| `set_up_sync`             | `destination_type`, `accounts` (both optional) | Authorise a destination, choose a target and accounts, create the sync, watch the first run |
| `monthly_spending_review` | `month` (optional, `YYYY-MM`)                  | Spending by category and account with current balances                                      |
| `fix_failing_sync`        | `sync` (optional)                              | Diagnose a sync that needs attention and apply the recommended action                       |
| `connect_bank`            | `institution` (optional)                       | Start bank consent and confirm the new connection and accounts                              |

## Guides

Before a multi-step task, the AI can call `redbark_guide` with a matching topic. Guides are condensed from the documentation:

* `onboarding`: connect a bank, authorise a destination, first sync.
* `syncs_and_destinations`: sync types, destinations, mappings, runs, and resync.
* `rules_and_categories`: rulesets, rules, preview, taxonomy, and source claims.
* `events_and_webhooks`: event destinations, events, redelivery, and signatures.
* `errors_and_scopes`: errors, request ids, scopes, and plan gates.
* `api_conventions`: public ids, lists, pagination, idempotency, and versioning.

For individual tool arguments, see [tools and arguments](/docs/mcp/tools). If a workflow fails, see [troubleshooting](/docs/mcp/troubleshooting).