Actual Budget

**Quick judgment** This skill is a powerful, high-coverage interface to self-hosted Actual Budget instances. It is well-suited for agents that need deep, programmatic access to a user’s personal finance data (budgets, accounts, transactions, rules, and analytics). However, it carries significant privacy and data-integrity risk because it enables write operations (imports, edits, rules, schedules, bank sync) on sensitive financial data. **What it does well** - Wraps the **official Actual Budget Node.js API**, covering: - Budget months and category allocations - Accounts (creation, closing, balances) - Transactions (query, import with deduplication, updates, splits, transfers) - Categories, payees, rules, schedules - Bank sync and full sync with the server - Advanced analytics via **ActualQL** (filtering, grouping, aggregation, search), including operators like `$lt`, `$gte`, `$like`, etc. - Uses a clear, consistent data model: - Amounts in integer cents - Dates as `YYYY-MM-DD`, months as `YYYY-MM` - UUID IDs with helpers to look up by human-friendly names. - Supports **local caching** via `ACTUAL_DATA_DIR`, improving responsiveness for repeated queries. **Key risks & safety considerations** - **High sensitivity of data**: Exposes detailed personal financial data (transactions, balances, categories, schedules). Any compromise or misuse could reveal highly private information. - **Destructive capabilities**: The skill can create/modify accounts, transactions, budgets, rules, and schedules, and run bank sync. An agent error or prompt-injection attack could: - Mis-categorize or delete transactions - Create incorrect transfers or accounts - Alter rules/schedules leading to ongoing misclassification - Trigger unintended bulk imports or sync operations - **Authentication & network exposure**: - Requires `ACTUAL_SERVER_URL`, `ACTUAL_PASSWORD`, and `ACTUAL_SYNC_ID` via environment variables. If these are mishandled, an attacker could gain access to the budget. - Self-signed certificate handling is supported via `NODE_EXTRA_CA_CERTS`, but if users circumvent TLS validation elsewhere, they risk man-in-the-middle attacks. - **No built-in fine-grained permission model in the described API**: An agent with this skill essentially has full control over the budget it can access. **Recommended scenarios** Use this skill when: - The user explicitly wants an **automation or analytical layer** on top of their self-hosted Actual Budget, such as: - A conversational agent to query spending, balances, and budget status. - Automated transaction import and categorization workflows. - Custom reports and ActualQL-based analytics for monthly reviews or audits. - Bulk migration scripts from other budgeting tools. - The environment can enforce **strong safeguards**, such as: - Network isolation and secure TLS configuration to the Actual server. - Secure handling of environment variables (passwords and sync IDs). - Policy-level restrictions on when write operations (imports, updates, rule creation, schedule creation) are allowed, ideally with explicit user confirmation. Avoid or heavily constrain this skill when: - You cannot guarantee secure storage of credentials and a trusted network path to the Actual Budget server. - The agent is allowed to execute arbitrary user-provided instructions without guardrails, especially around imports, bank sync, or rules/schedules. **Overall** This is a strong, feature-complete skill for Actual Budget power users and automation scenarios. It should be treated as **high-privilege** due to the breadth of read/write access to sensitive financial data, and deployed with strict confirmation flows and environment security.