Stele

About Stele

Stele is a shared project memory that every AI coding agent on a project reads before it acts and writes back to as it learns. It is a live ledger rather than a passive log — decisions, lessons, risks, and open tasks in one record, surfaced at the moment a change would repeat a past mistake. Three ideas hold it together. First, one living memory: architectural decisions and hard-won lessons live in the record, not in someone's head or a stale markdown file. Second, work the whole team can see: tasks belong to the project rather than a person, and agents claim them atomically so two sessions never start the same work, while a human operator can pick up the open queue and run it end to end. Third, continuity across tools: plan with one agent, build with another, switch when a quota fills, and Claude Code, Cursor, Codex, Antigravity, Copilot, OpenCode, Grok, and Kimi Code CLI all read the same knowledge graph — so the project's memory belongs to the team, not to whichever vendor you were using that week. Setup is three commands: one line installs the CLI, you sign in with your invited email, and you run /stele:start once inside your agent to wire it into the project and build the first record. After that it is automatic — you keep prompting exactly as you did before. It is invite-only beta as of July 2026, with a waitlist, and runs on macOS, Linux, and Windows from a single binary.

Key Features

Stele Pros & Cons

Who Is Stele Best For?

Tags