{"Status":true,"Message":"","Response":{"post":{"postuid":"1882ba48-a09a-431c-ac71-ffaa31ffc807","tenantuid":"d8b744fc-2e70-4089-bb80-dd1d08f6c7b2","projectuid":"fc6490ac-7527-4f49-b06e-46f701280e85","title":"Review Comments and AI Agents","slug":"article/review-comments-and-ai-agents","html":"\u003Cp\u003EReview comments let you leave instructions on an article for an AI agent, such as \u003Ci\u003E\u0022Update the stats to 2026\u0022\u003C/i\u003E. HelpGuides calls this Cowork. Any AI assistant connected to your project through \u003Ca href=\u0022/article/model-context-protocol\u0022\u003EModel Context Protocol\u003C/a\u003E can find open comments, make the edits, and mark each comment resolved with a note about what it changed. You check the result and publish it.\u003C/p\u003E\u003Ch2 id=\u0022leave_a_comment\u0022\u003ELeave a comment\u003C/h2\u003E\u003Col\u003E\u003Cli\u003EOpen the article in the editor and click \u003Cb\u003EAdvanced\u003C/b\u003E to open \u003Cb\u003EAdvanced Article Options\u003C/b\u003E.\u003C/li\u003E\u003Cli\u003EClick the \u003Cb\u003EReview\u003C/b\u003E tab.\u003C/li\u003E\u003Cli\u003EDescribe the change under \u003Cb\u003ENew comment\u003C/b\u003E.\u003C/li\u003E\u003Cli\u003EClick \u003Cb\u003EAdd Comment\u003C/b\u003E.\u003C/li\u003E\u003C/ol\u003E\u003Cfigure\u003E\u003Cimg class=\u0022lazy-load\u0022 data-src=\u0022https://graffiti-auf7e6dwhxhcbwek.z03.azurefd.net/d8b744fc-2e70-4089-bb80-dd1d08f6c7b2/fc6490ac-7527-4f49-b06e-46f701280e85/images/b71f350bdf2547868a4373e3ead7aa36.png?v=805553631\u0022 data-width=\u0022700\u0022 data-height=\u0022900\u0022 style=\u0022max-width: 100%;\u0022 alt=\u0022Review tab with the New comment box and Add Comment\u0022/\u003E\u003Cfigcaption\u003EThe Review tab\u003C/figcaption\u003E\u003C/figure\u003E\u003Cp\u003EThe \u003Cb\u003EReview\u003C/b\u003E tab shows a red badge with the number of open comments on the article. Open comments are picked up by any MCP-connected agent the next time it runs.\u003C/p\u003E\u003Ccite class=\u0022recommended\u0022\u003E\u003Cspan class=\u0022title\u0022\u003EWrite comments an agent can act on\u003C/span\u003E\u003Cp\u003EAsk for one change per comment and be specific: name the section, the facts to change and where the new facts come from.\u003C/p\u003E\u003C/cite\u003E\u003Ch2 id=\u0022comment_statuses\u0022\u003EComment statuses\u003C/h2\u003E\u003Ctable border=\u00221\u0022 style=\u0022border-collapse: collapse; width: 100%;\u0022\u003E\u003Cthead\u003E\u003Ctr\u003E\u003Cth\u003EStatus\u003C/th\u003E\u003Cth\u003EMeaning\u003C/th\u003E\u003C/tr\u003E\u003C/thead\u003E\u003Ctr\u003E\u003Ctd\u003EOpen\u003C/td\u003E\u003Ctd\u003EWaiting for someone to act on it. Only open comments count toward the badge.\u003C/td\u003E\u003C/tr\u003E\u003Ctr\u003E\u003Ctd\u003EResolved\u003C/td\u003E\u003Ctd\u003EThe requested edits were made. Shows the resolution notes and who resolved it (\u003Ci\u003EBy Cowork\u003C/i\u003E for an agent).\u003C/td\u003E\u003C/tr\u003E\u003Ctr\u003E\u003Ctd\u003EDismissed\u003C/td\u003E\u003Ctd\u003EClosed without edits, for example because the request was out of scope or a duplicate. Shows the reason, if one was given.\u003C/td\u003E\u003C/tr\u003E\u003C/table\u003E\u003Cp\u003EYou can close a comment yourself with the \u003Cb\u003EResolve\u003C/b\u003E or \u003Cb\u003EDismiss\u003C/b\u003E button under it.\u003C/p\u003E\u003Ch2 id=\u0022how_an_agent_works_through_comments\u0022\u003EHow an agent works through comments\u003C/h2\u003E\u003Cp\u003EConnect an assistant such as \u003Ca href=\u0022/article/adding-helpguides-to-claude-as-an-mcp-connector\u0022\u003EClaude\u003C/a\u003E or \u003Ca href=\u0022/article/add-chatgpt-as-an-mcp-connector\u0022\u003EChatGPT\u003C/a\u003E to your project, then ask it something like \u003Ci\u003E\u0022Work through the open review comments in my docs.\u0022\u003C/i\u003E It uses these tools:\u003C/p\u003E\u003Ctable border=\u00221\u0022 style=\u0022border-collapse: collapse; width: 100%;\u0022\u003E\u003Cthead\u003E\u003Ctr\u003E\u003Cth\u003ETool\u003C/th\u003E\u003Cth\u003EWhat it does\u003C/th\u003E\u003C/tr\u003E\u003C/thead\u003E\u003Ctr\u003E\u003Ctd\u003E\u003Ccode\u003Eget_articles_with_open_comments\u003C/code\u003E\u003C/td\u003E\u003Ctd\u003ELists articles that have open comments, with the number of open comments on each.\u003C/td\u003E\u003C/tr\u003E\u003Ctr\u003E\u003Ctd\u003E\u003Ccode\u003Elist_agent_comments\u003C/code\u003E\u003C/td\u003E\u003Ctd\u003ELists every open comment in the project with its article, oldest first.\u003C/td\u003E\u003C/tr\u003E\u003Ctr\u003E\u003Ctd\u003E\u003Ccode\u003Eget_comments\u003C/code\u003E\u003C/td\u003E\u003Ctd\u003EReturns all comments on one article, including resolved and dismissed ones, so the agent sees earlier instructions too.\u003C/td\u003E\u003C/tr\u003E\u003Ctr\u003E\u003Ctd\u003E\u003Ccode\u003Eresolve_comment\u003C/code\u003E\u003C/td\u003E\u003Ctd\u003EMarks a comment resolved, with optional \u003Ccode\u003Eresolution_notes\u003C/code\u003E that summarize the edits.\u003C/td\u003E\u003C/tr\u003E\u003Ctr\u003E\u003Ctd\u003E\u003Ccode\u003Edismiss_comment\u003C/code\u003E\u003C/td\u003E\u003Ctd\u003ECloses a comment without edits, with an optional reason.\u003C/td\u003E\u003C/tr\u003E\u003C/table\u003E\u003Cp\u003EA typical run: the agent lists the open comments, reads the article with \u003Ccode\u003Eget_article\u003C/code\u003E, edits it with \u003Ccode\u003Eupdate_article_blocks\u003C/code\u003E, then calls \u003Ccode\u003Eresolve_comment\u003C/code\u003E with a short note. Edits are saved as a working copy, so the live article doesn\u0027t change until you publish. See \u003Ca href=\u0022/model-context-protocol-endpoints\u0022\u003EModel Context Protocol Endpoints\u003C/a\u003E for every tool.\u003C/p\u003E\u003Ch2 id=\u0022get_a_slack_message_when_a_comment_is_resolved\u0022\u003EGet a Slack message when a comment is resolved\u003C/h2\u003E\u003Cp\u003EWith Slack connected, turn on \u003Cb\u003ENotify on Cowork updates\u003C/b\u003E in \u003Cb\u003ESettings\u003C/b\u003E on the \u003Cb\u003EIntegrations\u003C/b\u003E tab. Each time a comment is resolved, Slack gets a message that links to the article and shows the comment and the resolution notes, so someone can check the draft before it\u0027s published. Dismissed comments don\u0027t send a message. See \u003Ca href=\u0022/article/connecting-slack\u0022\u003EConnecting Slack\u003C/a\u003E.\u003C/p\u003E\u003Ch2 id=\u0022review_and_publish\u0022\u003EReview and publish\u003C/h2\u003E\u003Cp\u003EOpen the article to see the agent\u0027s working copy, check the changes against the comment\u0027s notes, and publish when you\u0027re happy. To give every agent standing guidance, such as tone or formatting rules, add \u003Cb\u003EPrompt Instructions\u003C/b\u003E in \u003Ca href=\u0022/article/ai-settings\u0022\u003EAI Settings\u003C/a\u003E.\u003C/p\u003E","publish_status":0,"post_type":"Article","authoruid":"3dde8c16-763a-4a2b-ae0b-1d8c50c62e3d","author":{"authoruid":"3dde8c16-763a-4a2b-ae0b-1d8c50c62e3d"},"featured_image_updating":false,"meta_description":"Leave actionable AI review comments in articles with Cowork via Model Context Protocol; agents edit, resolve, and notify you to review and publish efficiently.","display_toc":true,"has_workingcopy":false,"allow_indexing":true,"total_views":1,"date_published":"2026-10-04T21:47:31.137","date_updated":"2026-10-04T21:47:31.783","date_created":"2026-09-28T16:20:21.497"}}}