Some facts in a project are not beliefs. "Never deploy the payments service on a Friday" is not something the agent should re-derive each session, weigh against newer context, or quietly forget under budget pressure. It is a standing constraint, and until now daimon had no home for one: a rule stated once landed in the belief partition, did not carry, and was first in the drop order when the briefing ran over budget. 0.31.0 adds pinned rulings.

A memory that only records what you decided is half a memory. The other half is what you ruled out, and why, so nobody spends another afternoon rediscovering it. 0.29.0 adds that half.

Everyone building agent memory is building the same thing: remember what worked, remember what was decided, remember who prefers tabs. Useful. But the expensive amnesia is not forgetting successes. It is forgetting failures.

An agent that forgets a success re-derives it in a few hundred tokens. An agent that forgets a failure re-attempts it: the refactor that broke prod, the "obvious cleanup" that was load-bearing, the dependency upgrade that was abandoned twice for the same reason. You pay for the same dead end every time a fresh session walks into it confidently.

Here is a trap every agent memory system walks toward sooner or later: an item that keeps showing up starts to feel true. Five sessions all "remember" the same fact, so the fact must be solid. Promote it. Trust it more.

We wanted that feature. Independent re-observation should strengthen a memory — that is how evidence works everywhere else. But before building it we went looking for the ways it breaks, and what we found changed the design, the release, and one security assumption we had been living with. All of it shipped today in v0.22.0.

Your agent opens a session and tells you what you decided last time. Some of that is a quote. Some of it is the model's summary of a quote. Some of it is a conclusion the model drew at 2am from a session it was already losing track of.

All three arrive in the same font, in the same confident tone, with no way to tell them apart. That is the actual problem with agent memory, and it is not solved by remembering more.

daimon 0.19.0 is on PyPI, three days after 0.18. The headline is a pair of doors: the documentation site you are reading — in English and Spanish from day one — and a read-only MCP server, so hosts that speak MCP but have no hook system can still read your agent's memory. Also in the box: daimon forget ships, and a chunk cache stops failed serializes from re-buying work they already paid for.

daimon now has a documentation site — the one you are reading — in English and Spanish, with a quickstart that takes you from install to your first briefing, and concept pages for the ideas that make daimon different: trust classes, carry, receipts, and the item lifecycle. This blog is the new canonical home for releases, feature explainers, and field incidents; every announcement you see from us elsewhere will link back here.