Premortem
Premortem reads a Jira issue — and, if you connect it, the code written for that issue — and tells you how the work could go wrong before it does.
It is not a place to file risks. It is the thing that finds them, using your team’s own context: the epic, the architecture page, what people said in the comments, and the incidents you have already lived through. You accept the ones worth keeping; those go into the issue description and your Confluence risk register. The rest you dismiss and never see again.
It runs on your own AI key. Your text goes from your Atlassian site straight to the provider you chose — OpenAI, Anthropic, Google, or any OpenAI-compatible endpoint you point it at. We do not have a server in the middle, because we do not have a server at all.
Start here
- Getting started — install, add a key, first assessment. About five minutes.
- Settings reference — every option, what it costs, and when to change it.
Then, when you want more from it
- Code risk analysis — connect a Git repository so Premortem reviews the actual diff against what the team decided.
- The risk register — how accepted risks reach Confluence, and how the register is kept from rotting.
- Citing your own incidents — import past incidents and postmortems so assessments say “this is how PROJ-118 started” instead of speaking in generalities.
Before you decide
- Where your data goes — read this one before installing, not after.
- Troubleshooting — the failures people actually hit.
What it needs
| An API key from an LLM provider | Required. Nothing runs without it, and the tokens are billed to you |
| A Confluence space | Optional. Only if you want the risk register |
| A Git repository | Optional. Only for code risk analysis |
| Jira admin rights | To configure it. Anyone who can see an issue can then use it |
There is no account to create with us, and nothing to pay us for at this stage.