# SafeFetch

## Docs

- [Approve Action](https://docs.safefetch.dev/api-reference/actions/approve.md): Approves an action in `awaiting_approval` and moves it to `pending`.
- [Cancel Action](https://docs.safefetch.dev/api-reference/actions/cancel.md): Cancels actions in `pending` or `awaiting_approval` status.
- [Create Action](https://docs.safefetch.dev/api-reference/actions/create.md): Creates an action and returns the full enriched action object.
- [Get Action](https://docs.safefetch.dev/api-reference/actions/get.md)
- [List Actions](https://docs.safefetch.dev/api-reference/actions/list.md)
- [Retry Action](https://docs.safefetch.dev/api-reference/actions/retry.md): Creates a new pending action from a failed/cancelled action.
- [Authentication](https://docs.safefetch.dev/api-reference/authentication.md): Bearer API key authentication for all versioned endpoints.
- [API Overview](https://docs.safefetch.dev/api-reference/overview.md): Base URL, format conventions, limits, request IDs, and error model.
- [CLI Commands](https://docs.safefetch.dev/cli/commands.md): All SafeFetch CLI commands with usage examples.
- [CLI Install](https://docs.safefetch.dev/cli/install.md): Install and configure the SafeFetch CLI.
- [Actions](https://docs.safefetch.dev/concepts/actions.md): Understand action lifecycle, fields, available operations, and computed metadata.
- [HTTP Delivery](https://docs.safefetch.dev/concepts/http-delivery.md): Delivery semantics, signatures, retries, timeouts, and callback URLs.
- [Security](https://docs.safefetch.dev/concepts/security.md): How SafeFetch handles your data and what to consider when sending sensitive action bodies.
- [Admin Action](https://docs.safefetch.dev/guides/admin-action.md): Gate sensitive admin operations — user deactivation, permission changes, account deletions — behind a human approval step with a full audit trail.
- [CLI Reference](https://docs.safefetch.dev/guides/cli.md): Install and use the SafeFetch CLI to send, inspect, and manage actions from the terminal.
- [CRM Contact Update](https://docs.safefetch.dev/guides/crm-update.md): Safely update CRM contact records from AI agents — with deduplication to prevent double-writes and sync mode for immediate confirmation.
- [Deduplication and Idempotency](https://docs.safefetch.dev/guides/deduplication.md): Prevent duplicate work with dedupe keys and idempotency keys.
- [Human Approval](https://docs.safefetch.dev/guides/human-approval.md): Add a human decision gate before an action can execute.
- [MCP Server](https://docs.safefetch.dev/guides/mcp.md): Connect SafeFetch to Claude Desktop, Cursor, and other MCP clients
- [MCP Agents](https://docs.safefetch.dev/guides/mcp-agents.md): Use SafeFetch through MCP tools from AI agents.
- [Retry and Backoff](https://docs.safefetch.dev/guides/retry-backoff.md): Understand attempts, exponential backoff, and terminal failure handling.
- [TypeScript SDK](https://docs.safefetch.dev/guides/sdk.md): Use the safefetch npm package to send, inspect, and manage actions from TypeScript or JavaScript.
- [Stripe Refund](https://docs.safefetch.dev/guides/stripe-refund.md): Gate AI-initiated refunds behind a human approval step so no money moves without sign-off.
- [Introduction](https://docs.safefetch.dev/introduction.md): SafeFetch is a safety layer for AI actions. Give it a URL and a body. It makes the request with retries and stored results.
- [MCP Setup](https://docs.safefetch.dev/mcp/setup.md): Install and configure the SafeFetch MCP server for Claude Desktop.
- [MCP Tools](https://docs.safefetch.dev/mcp/tools.md): Parameters and behavior for all SafeFetch MCP tools.
- [Quickstart](https://docs.safefetch.dev/quickstart.md): Add a safety layer to your AI actions in minutes.

## OpenAPI Specs

- [openapi](https://docs.safefetch.dev/openapi.yaml)