Your agents don’t need to persist. Your purpose does.
How do I connect Sentry to ChatGPT?
The error stream over MCP reaches ChatGPT like any client — projects, issue detail, retrieval, Seer's fix flows.
Direct answer: connect Sentry's MCP server to ChatGPT. In ChatGPT, add it from the connector settings (developer mode); the tool's own docs carry the current path. Authorised, ChatGPT reads projects, issue details and errors, with Seer fix flows available — triage questions answered from the stream during the conversation that needs them.
Triage, wherever you think
What spiked since the deploy, who is affected, what the breadcrumbs say — the debugging read, in the assistant you happen to have open. The stream is the stream; the protocol makes it portable.
A fixed bug that returns jumps the queue
Worked case: your team ratified a regression rule — any issue marked resolved that reopens goes to the front of the queue, no re-litigation, because a twice-shipped bug once taught you that reopened means the mental model is wrong, not just the code. Sentry tracks resolved-then-reopened states cleanly; the queue-jump is your law about what that state means.
Through Unl the triage read arrives with the law standing: two reopened issues sitting mid-queue — front of the line, by your rule, the wrong-mental-model lesson attached. Prioritisation debates end where your own precedent already ruled.
Sentry to ChatGPT reads the error stream; through Unl the stream answers to the triage laws you ratified — the reopened bug claims its place because you already decided it should.
Reads through Unl arrive with measured context — in the presence of the decisions you’ve already settled. It works today: one box, paste anything. If it speaks MCP, Unl can reach it. Readings arrive unprompted, the data beside the criterion; Unl is a courier, not a warehouse, and keeps only your keys and the frame.
Read further
Questions people ask
What does the Sentry surface give ChatGPT?
Projects, issue details, error retrieval and Seer fix flows — the same observational and remediation surface every MCP client reads.
Why ratify triage rules instead of leaving them to judgement?
Because the judgement was already made — at the post-mortem. Ratifying it stops each new on-call rediscovering it under pressure.
Does Unl act on the issues?
Not while measuring — a read only reads, so checking the stream against your triage rules never changes an issue. Past that the honest answer is that it can: where a server exposes a tool that acts, Unl can call it, and that call carries your explicit in-turn gesture, checked before the call goes out, with anything not declared read-only treated as a write by default. So nothing is triaged, assigned or resolved as a side effect of a read — only when you ask and confirm.
Can I just use Sentry's own MCP server on its own?
Yes — for the raw read it is a strong surface, and this page's steps need nothing else. Projects, issue detail, retrieval and Seer's fix flows all come back accurately. What it cannot tell you is whether the error stream has crossed the line you drew, because the budget and the reason you set it live with you, not in the tracker. That is the layer Unl adds on top of the same read, not a replacement for it.
What this is
Your agents don’t need to persist. Your purpose does.
Save the thoughts, decisions and targets worth keeping, each with its reasoning, carried into every AI session the moment they matter. A new unit of exchange between you and your AI: the Settled Why with standing that travels. Unprompted.
Your whole AI world. What you decided at the epicentre. It works inside Claude, Claude Code, ChatGPT and Cursor, and connects in a couple of steps.
MCP native·Human settled·Model agnostic·Your data
Measured Context
Connect Unl to bring the right information into the moment.
Your sources, read against the criteria you set.
The full product, open. Free to start. Paid plans from $10 a month.