agents forget.
partyline doesn’t let them.
You decide something with Claude today. Tomorrow’s session re-litigates it. Your teammate’s Codex never heard about it at all. Context Threads keep one team memory — decisions, constraints, contracts — and every agent partyline runs starts already briefed.
$ brew install partyline-sh/tap/partylinehow it works
attach a thread
Any session partyline launches is wired in automatically. The agent gets the team’s current facts at launch — not a to-do list, just “here’s what we know.”
it remembers for you
Agents record decisions, constraints, and contracts as they work. Raw conversation never leaves your machine — only the distilled facts.
every agent starts briefed
Planners, builders, reviewers — a new machine, a different AI, a teammate's session — all read the same live memory.
what you're wondering
how is this not a CLAUDE.md?
A file is per-repo, per-tool, and nobody keeps it current. A thread is shared across people and tools, attributed, versioned, and updated as the work happens.
can a wrong fact steer an agent?
No. Agents only ever see confirmed, live facts. Retracted, superseded, and unreviewed entries are invisible to them — the history stays visible to you.
which tools?
The launch briefing works with every agent partyline runs. Live in-session recall is Claude Code today; Codex is experimental, more on the way.
stop re-explaining your project.
$ brew install partyline-sh/tap/partylinethe memory behind the pipeline — see the whole system →