> For the complete documentation index, see [llms.txt](https://docs.3techagency.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.3techagency.com/documentation/work-safely/approvals-undo-and-history.md). # Approvals, Undo and History Approval cards, Task proposals, Undo, rollback conflicts and the change history. ## Approval cards When a tool is classed **Ask** (deletes, bulk changes, code, file, database or user writes, broad edits), Chat pauses on an **Approval needed** card. It shows the tool, the target and the exact arguments. Before approving, check: * the target page, post, product or category * the scope: how many items, which category, which search text * numbers such as a sale percentage or dates * the integration that will do the write **Approve** runs that call. **Approve all** approves every pending call in the turn. Cancelling skips those changes ("Skipping those changes") and the assistant continues without them. Approving a **dry run** only previews; the real write can ask again. Protected pages are refused even with Approve all. To approve, a role needs **Approve writes** under Settings > Access. Administrators always can. ## Task proposals Tasks in **Preview changes first** mode stage their writes as **Proposed changes**. 1. Open the run from **Playbooks > Tasks and runs > Run history** or the link in Home or the email. 2. Select the rows you want. 3. Choose **Apply selected**. 4. Wait for the list to refresh before selecting more. If refreshing fails, reopen the run. Proposals not reviewed within **14 days** expire and can no longer be applied. A run that would change more than **25 items** switches to preview, and applying any selection needs a verified database backup made within the last hour (see [Backups and restore](/documentation/reference-and-help/backups-and-restore.md)). **Undo run** reverses every change that run made. ## Undo Undo is available from: * the **Undo** button on a Chat result card, and **Undo last change** * **Home > Recent changes** * **History**, with **Roll back** on each entry * **Undo run** on a Task run If the target changed after the AI's edit, rollback stops and warns: "This target changed since the change was recorded. Rolling back now would overwrite the newer edits." Choose **Roll back anyway** only if you want to lose those newer edits. If a large before-image was capped, you see "Change partially rolled back". Undo is not universal. Changes made by other plugins, external services, uploaded files and changes made outside WP AI Manager may show **Undo unavailable**. An Undo can itself require approval. ## History **WP AI Manager > History** records every AI change to content, page builders, settings, files and the database. * **Filters** are built from the domains on your site: Content, Elementor, Page builders, SEO, WooCommerce, Settings, Media, Redirects, Filesystem, Database. * **Who** shows the user, or **System** for scheduled runs and WP-CLI. * **Delete** removes one entry. Deleting an entry that can still be rolled back discards that ability permanently. * **Clear all** empties the history and cannot be undone. * **Backups** lists Task backups; see [Backups and restore](/documentation/reference-and-help/backups-and-restore.md). ## Recovery practice For anything important, keep a separate full-site backup with your backup provider. WP AI Manager's own archives and large-write checkpoints are **database-only**. Test a restore on a disposable copy of the site first. --- # 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 by asking a question. Perform an HTTP GET request on the following URL with the `ask` and `goal` query parameters: ``` GET https://docs.3techagency.com/documentation/work-safely/approvals-undo-and-history.md?ask=&goal= ``` `ask` is the immediate question: it should be specific, self-contained, and written in natural language. `goal` is what the user is ultimately trying to achieve, the reason they need the answer. Sharing it helps GitBook give you a better, more relevant answer. A goal is most helpful when it describes the outcome the user wants rather than restating the question. For example, with `ask=how do I create an API token`, a goal like `automate deployments from our CI pipeline` lets GitBook tailor the answer to that use case. 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.