How much does Sentry cost?

Seer is $40 per active contributor per month, added to a paid plan. An active contributor is anyone who makes two or more pull requests in a month to a repository with a Seer feature turned on, and counts reset each month. Base plans: Developer is free for one user and has no Seer, Team is $29 a month or $26 billed annually, Business is $89 a month or $80 billed annually, and Enterprise is custom. New accounts get a 14 day Business trial, Seer has its own one time trial, and Seer is billed apart from the pay as you go budget.

Does Sentry support MCP and tool calling?

Full under the Agentic Index verification standard. Sentry publishes public evidence of Model Context Protocol (MCP) support, tool calling, and API or SDK extensibility for connecting agents to external tools. Evidence basis: A REST API reaches the platform on Team plans and above, with region specific domains for US and EU organizations. A hosted MCP server at mcp.sentry.dev signs in with OAuth and can be scoped to one organization or one project, letting Claud. Graded from public evidence only against the Agentic Index 14 point capability taxonomy. No vendor pays for placement.

Does Sentry have SOC 2, SSO and role based access control?

Full under the Agentic Index verification standard. Sentry publishes public evidence of enterprise security and identity controls, covering a compliance posture such as SOC 2 or ISO 27001, single sign on (SSO or SAML), and role based access control (RBAC). Evidence basis: Access runs on organization roles, Billing, Member, Manager and Owner plus a legacy Admin, with team level Contributor and Admin roles on Business and Enterprise. SAML 2.0 single sign on and SCIM provisioning come with Business and Enterpri. Graded from public evidence only against the Agentic Index 14 point capability taxonomy. No vendor pays for placement.

Can Sentry be self hosted or run in a private VPC?

Full under the Agentic Index verification standard. Sentry publishes public evidence of self hosted or private VPC deployment, along with data residency options for controlling where agent data is processed and stored. Evidence basis: An organization picks its data storage location when it is created: the United States, hosted in Iowa, or the European Union, hosted in Frankfurt, with region specific API domains at us.sentry.io and de.sentry.io. Events, traces, logs, prof. Graded from public evidence only against the Agentic Index 14 point capability taxonomy. No vendor pays for placement.

Back to vendors

Sentry

Also known as: Sentry Seer, Seer by Sentry, Sentry Autofix, Sentry AI Code Review, sentry.io, Functional Software, Inc.

Visit site

SRE / DevOps agentindependentVerified 2026-11-08

Entry priceSeer is $40 a month per active contributor, added to a Team plan at $29 a month or $26 billed annually, or a Business plan at $89 a month or $80 billed annually. Enterprise is custom priced.Full pricing detail

Sentry, the application monitoring platform software teams use to track errors, traces, logs and replays, sells Seer, an AI debugging agent that works from that production data. Seer finds the root cause of new issues on its own, drafts a fix and opens a pull request, reviews pull requests for errors Sentry has already seen, and answers plain language questions about production in Sentry or Slack. It can also hand a fix to Claude Code, Cursor or GitHub Copilot.

Sentry sells application monitoring for software teams: error monitoring, tracing, structured logs, session replay, profiling, uptime and cron monitors, and monitoring for the AI agents and MCP servers customers run themselves. Developers send telemetry through open source SDKs for web, mobile, backend and desktop platforms, and Sentry groups it into issues tied to releases, commits and code owners. Seer, its AI debugging agent, works on top of that data and is sold as an extra subscription on any paid plan.

Seer has three modes. Autofix picks up new production issues on its own once they pass a fixability check, traces the root cause through stack traces, spans, logs, profiles and the linked GitHub or GitLab repositories, proposes a plan a developer can edit, then writes the fix and opens a pull request. It keeps pushing commits until CI passes and answers reviewer comments, or it hands the root cause and plan to Claude Code, a Cursor Cloud Agent or a GitHub Copilot cloud agent to write the code instead.

Code Review checks each pull request for errors Sentry has already seen in production and reports through comments and a GitHub status check. Seer Agent, in open beta, answers plain language questions about production data, pulls in Datadog, Google Cloud and Grafana telemetry, and can archive issues or build dashboards, monitors and alerts from the conversation, in Sentry or in Slack.

Seer costs $40 a month per active contributor, meaning anyone who opens two or more pull requests in a month in a connected repository, on top of a Team plan at $26 a month billed annually or a Business plan at $80. The free Developer plan does not include it. Organizations choose US or EU data storage when they sign up, Seer's models run inside Sentry's own production infrastructure with Anthropic and OpenAI as AI subprocessors, and customer data is not used for training unless the organization opts in. Sentry holds SOC 2 Type II and ISO 27001, and Business adds SAML single sign on, SCIM and a BAA.

It fits engineering teams already sending errors and traces to Sentry who want fixes drafted from that evidence rather than from code alone, and teams that want pull request review informed by what actually breaks in production. It fits less well for teams on Bitbucket, Azure DevOps or self hosted GitHub and GitLab, which Seer does not support, for organizations that keep Open Membership off and so lose Seer Agent, and for buyers who want an incident agent that works across a wider monitoring stack without Sentry at the center.

Vendor details

Canonical URL

https://sentry.io

Subcategory

AI debugging agent that finds the root cause of production issues from monitoring data, opens fix pull requests and reviews code before merge

Funding status

Private company. Functional Software, Inc., doing business as Sentry, based in San Francisco, with affiliates in Vienna, Toronto and the Netherlands.

Company status

independent

Use cases & customers

Primary use cases

root cause analysis of production errors and performance issuesdrafting fixes and opening pull requests for new issuescode review that predicts errors from production historyplain language investigation of production telemetry in Sentry or Slackhanding fixes to Claude Code, Cursor or GitHub Copilot

Target customers

engineering teams already using Sentry for error monitoring and tracingdevelopers working on GitHub.com or GitLab.complatform and on call teams investigating production issuesteams adopting coding agents such as Claude Code, Cursor or GitHub Copilot

Deployment options

SaaS with US or EU data storage

Integrations

Seer connects to GitHub.com and GitLab.com through Sentry's own apps to read code, open pull requests and post reviews, and hands fixes to Claude Code, Cursor Cloud Agents and GitHub Copilot cloud agents. Seer Agent reads Datadog, Google Cloud and Grafana data and runs in Slack. The wider platform carries an integration catalog for source control, issue tracking and alerting, open source SDKs for most languages and frameworks, a REST API, and a hosted MCP server that connects Sentry to Claude Code, Cursor and VS Code.

In practice

A checkout service starts throwing a null reference error overnight. Once the issue passes ten events, Seer traces it to a commit that changed a payment response, drafts a guard and a regression test, and by morning a pull request is waiting with CI passing.

A developer marks a pull request ready for review. Seer flags that a renamed field is still read by a background job that failed in production last month, and comments on the exact line before anyone merges.

An on call engineer asks Seer Agent in Slack why checkout latency rose this week. Seer pulls traces from Sentry and metrics from Datadog, points to a slow database query, and creates an alert when the engineer asks for one.

Sources & related URLs

Related / legacy domains

https://docs.sentry.io https://mcp.sentry.dev https://changelog.sentry.io

Research sources

https://sentry.io/product/seer/ https://docs.sentry.io/product/ai-in-sentry/seer/ https://docs.sentry.io/product/ai-in-sentry/seer/autofix/ https://docs.sentry.io/product/ai-in-sentry/seer/code-review/ https://docs.sentry.io/product/ai-in-sentry/ai-privacy-and-security/ https://docs.sentry.io/organization/data-storage-location/ https://docs.sentry.io/organization/membership/ https://docs.sentry.io/pricing/ https://sentry.io/pricing/ https://sentry.io/security/ https://sentry.io/legal/subprocessors/ https://sentry.io/terms/ https://mcp.sentry.dev

Agentic Index coverage score

10.5 / 14 capabilities · 75%

Integrations & Tool Calling	Full
Seer works inside the developer's own systems through the Sentry GitHub and GitLab apps. It pushes branches and opens pull requests or merge requests with its fixes, updates those pull requests when CI checks fail, and posts review comments and status checks. At the code step it can pass the root cause and plan to Claude Code, a Cursor Cloud Agent or a GitHub Copilot cloud agent instead. Seer Agent reads metrics, traces and logs from Datadog, Google Cloud and Grafana, and runs inside Slack. Sourcedocs.sentry.io/product/ai-in-sentry/seer/autofixread 2026-11-08
Workflow Orchestration	Full
Every fix runs as a fixed sequence of root cause analysis, a solution plan, code generation and then iteration on the pull request. Teams set how far automation goes for each project, stopping after the root cause, after the plan or once a pull request is drafted. The code step can pass to an outside coding agent, which receives the root cause and plan as a structured prompt, and one fix can span several repositories when the cause sits in a dependent service. Seer then keeps pushing commits until CI passes, within a cap on automatic attempts. Sourcedocs.sentry.io/product/ai-in-sentry/seer/autofixread 2026-11-08
Knowledge Grounding & RAG	Partial
Each investigation draws on the organization's own production record: error events and stack traces, distributed traces and spans, structured logs, profiles and the code in linked repositories, including several repositories for distributed services. Seer also reads Sentry's own documentation and, through separate integrations, Datadog, Google Cloud and Grafana telemetry. That context is gathered fresh for each run. The product has no knowledge base, document index or wiki connector for runbooks and design notes that Seer searches across investigations. Sourcedocs.sentry.io/product/ai-in-sentry/seerread 2026-11-08
Human Oversight & Guardrails	Full
People decide how far Seer goes. Each project's automation setting stops it after the root cause, after the solution plan, or once a pull request is drafted, whatever the fixability score says. Before any code is written, the plan appears as a list of steps that a developer can trim or extend before choosing Code It Up. Organization admins can switch off code generation entirely, which removes the Create PR button, and can turn automatic CI iteration off project by project. When Seer reaches a gap in its understanding it stops and asks. Sourcedocs.sentry.io/product/ai-in-sentry/seer/autofixread 2026-11-08
Security, Identity & Governance	Full
Access runs on organization roles, Billing, Member, Manager and Owner plus a legacy Admin, with team level Contributor and Admin roles on Business and Enterprise. SAML 2.0 single sign on and SCIM provisioning come with Business and Enterprise, and Google and GitHub sign in with Team. Sentry holds SOC 2 Type II and ISO 27001 and signs a BAA on Business plans and above. Seer Agent queries data across the whole organization with no finer controls yet, so it is switched off wherever Open Membership is turned off. Sourcedocs.sentry.io/organization/membershipread 2026-11-08
Observability & Auditability	Full
Every Autofix run streams Seer's reasoning live and ends with a root cause write up that lays out the sequence of events, with annotations linking each step to the exact lines of code and the Sentry events behind it. The solution plan, the proposed diffs across repositories and the feedback given along the way stay on the issue. Seer Agent conversations are kept and can be reopened or shared, code review outcomes land as GitHub status checks, and the organization keeps its own audit log. Sourcedocs.sentry.io/product/ai-in-sentry/seer/autofixread 2026-11-08
Memory & State Persistence	Partial
Seer Agent conversations are saved, so a team member can reopen an earlier investigation and pick it up, share it with colleagues or copy it into another agent. Within an Autofix run, Seer keeps the feedback and instructions people add at each step and applies them to the next one. Rules files written for Cursor, Windsurf, Cline and Claude Code are read as standing instructions. Nothing carries lessons from one issue to the next, and Seer has no memory store that it writes to and reads back across runs. Sourcedocs.sentry.io/product/ai-in-sentry/seerread 2026-11-08
Deployment & Data Residency	Full
An organization picks its data storage location when it is created: the United States, hosted in Iowa, or the European Union, hosted in Frankfurt, with region specific API domains at us.sentry.io and de.sentry.io. Events, traces, logs, profiles, replays and source maps stay in that region, while account, audit log, SSO and integration metadata sit in the US, and the choice cannot be changed later. Seer runs only on Sentry's cloud with GitHub.com and GitLab.com, never on self hosted Sentry, and its language models run inside Sentry's production infrastructure. Sourcedocs.sentry.io/organization/data-storage-locationread 2026-11-08
Prebuilt Agents / Templates / Packs	Partial
Seer is one agent with three ways in: Autofix for production issues, Code Review for pull requests, and Seer Agent for plain language investigation and changes inside Sentry, each switched on by project or repository. Sentry adds cookbook recipes on top, among them fixing issues from Slack, a weekly performance triage run by Claude Code through the MCP server, and a self healing workflow. There is no catalog of prebuilt agents or templates to install for other jobs, such as incident command or on call runbooks. Sourcesentry.io/product/seerread 2026-11-08
Triggers & Channel Coverage	Full
Production events start most of the work. A new issue with ten or more events, seen within the last 14 days and with a high enough fixability score, launches Autofix with nobody asking. Code review runs on GitHub pull request webhooks when a pull request opens, moves to ready for review or gets a new commit, and on GitLab when a merge request opens or updates. People can also start it with an @sentry review comment, Ask Seer inside Sentry, Seer Agent in Slack, a Fix with Seer button on Slack alerts for early adopter organizations, or the Sentry MCP server from an editor. Sourcedocs.sentry.io/product/ai-in-sentry/seer/code-reviewread 2026-11-08
Model Flexibility & Routing	Partial
Sentry picks the models. Seer runs on third party large language models hosted inside Sentry's own production infrastructure, out of reach of the model makers, and Anthropic and OpenAI are both Sentry subprocessors for AI and ML services. Customers cannot choose a model or provider or bring their own key, and no setting routes work between models. Customer data is not used for training unless the organization opts in, and admins can switch off every generative AI feature at once. Sourcesentry.io/legal/subprocessorsread 2026-11-08
APIs / SDKs / MCP Extensibility	Full
A REST API reaches the platform on Team plans and above, with region specific domains for US and EU organizations. A hosted MCP server at mcp.sentry.dev signs in with OAuth and can be scoped to one organization or one project, letting Claude Code, Cursor, VS Code or any MCP client search issues, read traces and logs, manage projects and start a Seer root cause analysis. Open source SDKs cover Python, JavaScript, React, Next.js, Go, Ruby, PHP, .NET and many other platforms. Sourcemcp.sentry.devread 2026-11-08
Testing, Debugging & Optimization	Partial
Seer checks its pull requests against the customer's own CI. When checks fail it reads the failures and pushes follow up commits, within a limit on automatic attempts, and it also acts on review comments from people and from review bots. Code Review predicts errors in each pull request from production history and reports through a GitHub status check. A fixability score from a Sentry model decides which issues get automatic runs. Seer's fixes are never scored against saved test cases, and agent versions are not compared before rollout. Sourcedocs.sentry.io/product/ai-in-sentry/seer/autofixread 2026-11-08
Browser / Computer-use	Not documented
Seer works through telemetry, repositories and APIs. It reads stack traces, spans, logs and code, and it acts through GitHub, GitLab, Slack and Sentry itself, with no hosted browser and no control of a desktop or a web interface. Session Replay records how real users moved through the customer's web app, and Seer reads those recordings as evidence when it looks for a root cause, without ever driving the app or clicking through its pages on its own. Sourcedocs.sentry.io/product/ai-in-sentry/seerread 2026-11-08

The Agentic Index coverage score grades every vendor Full, Partial or Not documented against the same 14 buyer facing capabilities, from public evidence only. Each capability links to how all vendors in the index score on it. How this evidence is graded

Pricing

Seer is $40 a month per active contributor, added to a Team plan at $29 a month or $26 billed annually, or a Business plan at $89 a month or $80 billed annually. Enterprise is custom priced.

Seer is billed per active contributor per month, on top of a flat monthly or annual base plan that includes set volumes of errors, spans, logs and replays, with a pay as you go budget for more data.

Trial available

Included quota

Paid plans include 50,000 errors, 5GB of logs, 5GB of application metrics, 5 million spans, 50 replays, one uptime monitor, one cron monitor and 1GB of attachments each month. Seer use is not metered per run under current pricing; every active contributor in a repository with a Seer feature turned on is billed once a month.

What is public

Base plan prices for Team and Business, monthly and annual, the included event volumes, per category overage rates, the Seer price per active contributor and how contributors are counted, and the trial terms.

Billing mechanics

Seer is added to an existing subscription from the billing settings and appears as its own monthly charge, apart from the pay as you go budget. Base plans bill monthly or annually in advance, with pay as you go usage billed monthly in arrears.

Cost watchouts

Turning on Code Review for a single repository starts contributor billing. On GitHub every repository member who opens two or more pull requests counts, except accounts marked as bots; on GitLab every group member account counts, including bots and accounts used for Autofix. Seer needs a paid base plan on top.

Variable cost rationale

The Seer charge grows with the number of people opening pull requests in connected repositories, not with how many issues Seer investigates or how many reviews it runs, so a busy month for a small team costs the same as a quiet one. The base plan is flat, and event data beyond the included volume draws on a pay as you go budget that the customer caps.

Additional watchouts

Contributor counts include anyone who opens two or more pull requests in a connected repository, not only the people who use Seer, and GitLab counts every group member account. Plan downgrades and cancellations take effect at the end of the contract term with no refunds.

Overage / add-ons

Seer has no per run overage. Event data above the included volume draws on the pay as you go budget at per category rates, and data sent after the budget runs out is dropped rather than billed.

Sales call required

Mixed (some tiers require a call)

Free / trial

New accounts get a 14 day Business trial, and Seer has a one time trial started from the subscription page. The free Developer plan does not include Seer.

Lowest paid plan

Team at $29 a month, or $26 a month billed annually, plus Seer at $40 per active contributor

Commercial notes

Sentry is sold by Functional Software, Inc., doing business as Sentry, of San Francisco. Seer moved from a per subscription fee with per run charges to per contributor pricing in January 2026.

Key ambiguities

Older subscribers may still be on legacy Seer pricing of $20 a month per subscription plus per run charges for issue scans and fixes, which is no longer offered to new customers.

Cancellation / refund

Plan upgrades take effect immediately. Downgrades and cancellations apply at the end of the current contract term and are not refunded.

Support SLA / resale

Community and email support on every plan. Enterprise adds a technical account manager and dedicated customer support.

Missing data

Enterprise pricing and any volume discount on Seer contributors.

Agentic Index verified 2026-11-08

Official pricing page Docs

Data confidence: high