Use Sentry with AI via MCP
Sentry collects errors and performance anomalies from running applications – and the list usually grows faster than teams can work through it. The official Sentry MCP server lets CompanyGPT evaluate and interpret that data directly.
What is Sentry?
Sentry is a platform for application monitoring and error diagnostics. Applications report exceptions, stack traces and performance data to Sentry, where they are grouped into issues, linked to releases and commits, and rated by frequency and number of affected users. Sentry’s source code is publicly viewable – though under the Fair Source License (FSL) rather than as OSI-approved open source. Sentry can be used as a hosted service or run on your own infrastructure.
For AI agents, the appeal lies in the diagnostic work: the raw data is extensive and technical, while the actual question is usually simple – since when has an error occurred, how many users does it affect, is it connected to the latest release. Via MCP, an agent can pull the relevant issues and turn them into an assessment people can act on.
Connection at a Glance
- MCP server
- Official (Sentry), remote – self-hosting possible
- Transport
- Streamable HTTP
- Authentication
- OAuth (remote); access token for self-hosted operation
- Operating model
- SaaS or a self-operated Sentry instance
- Permissions
- Organization and project rights of the connected account
How to Connect Sentry to CompanyGPT
Sentry operates an official, publicly usable MCP server at mcp.sentry.dev/mcp, addressed via streamable HTTP. CompanyGPT connects to this endpoint directly, with no additional layer in between.
Authentication runs via OAuth: users sign in with their Sentry account, and the agent works within the organizations and projects that account is entitled to. The team and project assignments maintained in Sentry therefore keep applying unchanged.
If you run Sentry yourself or want full control over the data path, the MCP server can also be run self-hosted inside your CompanyGPT environment and pointed at your own Sentry instance. In that variant, authentication uses a stored access token rather than OAuth. In both variants, your CompanyGPT administrators decide by role and group which teams get access to the integration – typically engineering, operations and technical support.
Typical Use Cases
How AI agents in CompanyGPT work with Sentry
Assess errors after a release
After a deployment, the team asks which new issues appeared and how many users are affected – a solid basis for deciding on a rollback.
Back up a support report technically
Support checks whether error entries match a customer report and hands engineering a ticket with dependable technical detail.
Weekly error overview
Operations has the most frequent and the newly appearing issues of a week summarized instead of walking through dashboards manually.
Frequently Asked Questions
Sentry and AI – the most important answers
Does Sentry have an official MCP server?
Yes. Sentry provides an official, publicly usable MCP server at mcp.sentry.dev/mcp, addressed via streamable HTTP.
Can we run the MCP server ourselves?
Yes. The Sentry MCP server can also be self-hosted and pointed at your own Sentry instance from within your CompanyGPT environment – a good fit if you already host Sentry yourself. Authentication then uses an access token instead of OAuth.
Which projects does the assistant see?
Exactly those the Sentry account connected via OAuth is entitled to. The existing organization and project structure in Sentry remains authoritative.
Does error data end up with an AI vendor?
CompanyGPT runs in your own cloud tenant and retrieves the data directly through the Sentry endpoint. Which models may process it and how log data is handled is configured in your instance.
Can the agent also triage or assign issues?
The scope follows the tools exposed by the Sentry MCP server and the account rights. In CompanyGPT an agent can be limited to read-only analysis when status changes should stay with the team.
Sentry and Your AI – GDPR-Compliant with CompanyGPT
CompanyGPT is the enterprise AI assistant in your own cloud tenant: all relevant AI models, your company data and integrations like Sentry – centrally managed, with roles and permissions. We set up the connection together with you.
Sentry is a trademark of its respective owner (Sentry (Functional Software, Inc.)). Its mention describes compatibility and does not imply any partnership or endorsement. Availability and scope of the connection depend on the respective vendor.