Expand description
Fencing for text that Tracker users wrote.
Issue descriptions and comments are the one part of a Tracker response that
an outsider fully controls, which makes them the injection surface an agent
actually faces (docs/adr/0001-security-model.md). We do not try to sanitise
that text — rewriting someone’s issue would be worse than useless. We label
its boundaries so the reader, human or model, can tell content from instruction.