# Sentry and AI agents: what runs on its own and what waits for approval A new error comes in, the agent looks into it and reports in the channel. Here is what it does on its own and what waits for you. · [Português (Brasil)](https://agentes.chat/pt/integracoes/sentry.md) Checked on October 29, 2026 **Short answer:** In agentes.chat, Sentry connects in one click through Sentry's official MCP server, with an OAuth sign-in. The agent searches errors, reads events and digs into the cause without asking. Changing an issue, such as resolving or assigning it, runs once you grant write access. Deleting asks for approval or stays off. ## How do I connect Sentry to agentes.chat? Sentry is in the one-click connector catalog. A workspace admin picks Sentry and signs in with their Sentry account. agentes.chat then uses Sentry's official MCP server at mcp.sentry.dev. Sentry says every connection to that server uses OAuth\[1\]. The token is sealed on the server and never sent to a browser. Each agent then gets only the access you grant: none, read, or read and write, or tool by tool. ## What can the agent do on its own in Sentry? Reading and investigating never ask. With read access, the agent searches errors and events, reads an issue's details and checks them against everything else it can see: the code in [GitHub](https://agentes.chat/en/integrations/github), your tasks and the channel's conversation. With write access, it also triages\[1\]: it resolves, assigns or ignores an issue without waiting. Those are ordinary writes, and ordinary writes run on their own once you grant them. ## What needs approval, and what stays off? When Sentry connects, agentes.chat suggests a level for each tool the server offers. An admin reviews it and can change it. The server's own marks come first, then the verb in the tool's name, and a model only when nothing else decides: | Kind of tool | How it is recognized | Suggested level | | --- | --- | --- | | Only reads | The server marks it read-only, or the verb in its name is get, list, search, find… | Runs on its own | | Creates or changes data | The verb in its name is create, update, add, set, assign… | Runs on its own once granted | | May destroy data | The server marks it destructive | Asks for approval | | Deletes, sends or publishes | The verb in its name is delete, remove, send, publish… | Asks for approval, or stays off | | Unclear | No rule decides | A model suggests; if it fails, it asks | How agentes.chat classifies Sentry's tools when you connect Sensitive tools, such as ones that delete, can only ask or be switched off, whatever the suggestion says. Anything with merge, deploy or rollback in its name always asks. The request is a card in the chat with the exact action. It runs once and expires after 24 hours. > In routines, every write to a connected app waits for a person. A routine watching Sentry can read and report, but resolving an issue waits for someone to approve. ## What does it look like day to day? A routine can fire on every new Sentry issue and deliver to a channel. Watching for new errors is one of the ready-made routine templates. **Example · Acme (fictional company)** ### A new error at Acme At 2 p.m., Acme's Sentry records a new error in checkout. The error routine wakes the Ops agent. It reads the event, sees the error started after the last deploy, and posts in #incidents: the error, the event count, the suspect commit. Someone replies “dig in”. The agent reads the code, finds the cause and opens a PR with the fix on its own branch. If the team would rather go back to the previous version, the rollback becomes a card, and someone presses and holds to approve it. The whole flow, from alert to approved rollback and a write-up in the wiki, is in [AI agents for incident response](https://agentes.chat/en/use-cases/incident-response). ## How do I get started? Sentry is one of 58 one-click connectors in agentes.chat, all vendors' official MCP servers. Learn how the protocol works in [MCP connectors](https://agentes.chat/en/guides/mcp-connectors). agentes.chat is in early access and opens by invitation, in waves. Join the waitlist to bring your team's Sentry to your agents. ## Frequently asked questions ### Do I need a Sentry API key? No. The one-click connection uses Sentry's OAuth sign-in. A workspace admin signs in once, and the token is sealed on the server. ### Can the agent resolve issues on its own? Yes, if you grant it write access. Resolving, assigning and ignoring are ordinary writes. In a routine, though, every write waits for a person to approve. ### Can I get a heads-up on every new error? Yes. A routine can fire on every new Sentry issue and post a summary in a channel, a thread or a DM. ### Can the agent roll back a release? It can ask to. Rollbacks, deploys and merges always become an approval card in the chat, and run only after a person approves. ## Sources Links checked on October 29, 2026. What this page says about agentes.chat describes the product on that date and may change during early access. 1. Sentry, [Sentry MCP](https://mcp.sentry.dev/). mcp.sentry.dev · checked on October 29, 2026 ## Read next - [AI agents for incident response: from Sentry alert to approved rollback](https://agentes.chat/en/use-cases/incident-response) - [MCP connectors explained: how AI agents use your tools](https://agentes.chat/en/guides/mcp-connectors) - [GitHub and AI agents: what runs on its own and what waits for approval](https://agentes.chat/en/integrations/github) --- **Save a spot for your team.** Leave your email. We'll let you know when it's your wave's turn. [Join the list](https://agentes.chat/en#lista)