Self-hosting error tracking is usually about one of two things: a compliance rule that says telemetry stays inside the network, or a bill that grew faster than the team. GlitchTip and self-hosted Sentry answer that directly. Parsemend answers a narrower version of it, keeping source code off someone else's servers, without asking you to operate the tracker.
What you actually have to operate
| Sentry (self-hosted) | GlitchTip | Parsemend | |
|---|---|---|---|
| Deployment | You run it | You run it | Hosted by Parsemend |
| What you operate | Many services, incl. Kafka and ClickHouse | PostgreSQL 14+ and one service; Valkey optional | Nothing |
| Open-source Sentry SDKs work | Yes | Yes | Yes (verified: PHP, Node) |
| Tracing | Yes | Yes (performance monitoring) | Yes |
| Session replay and profiling | Yes | Not listed | No |
| AI agent | No (Seer is unavailable) | Not listed | Your own, over MCP, reading code on your machine |
Self-hosted Sentry
You get the product, including tracing and session replay. You also get the operational surface: the install runs databases, message brokers and other services, Kafka and ClickHouse among them, and the failure modes are the failure modes of a distributed system. Teams who run it successfully usually have someone whose job is partly to run it.
The sharp edge that surprises people: Sentry's self-hosted docs list "Seer and other AI & ML features" as unavailable, because they are closed source. If you self-host to keep source code and telemetry inside your network, the AI agent is precisely the feature you cannot have.
GlitchTip
GlitchTip implements Sentry's ingest protocol and needs PostgreSQL 14 or later plus a single service, or separate web and worker services once you scale. Valkey or Redis is optional; without it, Postgres handles the task queue, cache and sessions. Its site lists error tracking, performance monitoring, uptime monitoring and logs.
If your requirement is "error tracking, inside our network, that the open-source SDKs can talk to", GlitchTip is the smallest correct answer and you should probably use it.
Parsemend
Parsemend is an error tracker your coding agent can read. It is a hosted service: you install the open-source SDK for your language, paste your project's DSN, and events arrive on Parsemend's servers. It stores errors, traces, logs and cron check-ins, runs uptime checks, and alerts by email, Slack, Discord or signed webhook.
What earns it a place on this page is where the agent runs. On the Developer plan, the CLI runs an MCP server on your machine over stdio. Your agent (Claude Code, Codex, Cursor or any MCP client) calls get_issue_context for the stack trace, events, trace and logs, then reads and edits the code in your checkout with its own tools. Parsemend never receives your repository, needs no GitHub access, and asks for no AI provider key. The model is whichever one your agent already uses, under whatever agreement you already have with its provider. The one thing that comes back is post_back: the git diff and a summary, sent after the agent asks you.
So the source and the inference stay where they are today. Self-hosted Sentry has no AI agent in the distribution, and GlitchTip lists none.
The honest limitation
Parsemend cannot be self-hosted. Your events, which can carry personal data if your application puts it there, live on Parsemend's servers. If your compliance rule is that telemetry never leaves the network, Parsemend does not satisfy it and GlitchTip does.
It also has no session replay and no profiling. Self-hosted Sentry has both. If those are why you self-host, this is not the tool.
Competitor details on this page were checked against glitchtip.com, GlitchTip's install docs and Sentry's self-hosted docs in October 2026. If something is out of date, mail corrections@parsemend.com.