Agentero
에이전트를 연결해 로컬 Vault에 연구 컨텍스트를 보관·Zotero에서 무결점 이동·읽고 이해·재활용
개인 계정의 공개 저장소 소유자는 GitHub로 로그인해 운영자를 확인하고 소개를 직접 관리할 수 있습니다. 관리자 승인은 필요하지 않습니다.
로그인한 뒤 확인 파일을 만들어 이 저장소의 기본 브랜치에 30분 안에 커밋해 주세요. 다른 브랜치나 포크의 파일은 인정하지 않습니다. 파일을 올린 뒤 아래에서 확인하면 관리자 승인 없이 관리할 수 있습니다. 자격을 유지하려면 파일을 남겨 두세요.
로그인이 필요하면 로그인 화면으로 이동합니다. 로그인 후 이곳으로 돌아와 원하는 확인 버튼을 다시 눌러 주세요.
프로젝트 폴더에서 /nomorevibe verify를 실행하면 소유권을 확인하고 소개를 직접 관리할 수 있습니다.
1분 만에 확인하기Jori: the open-source AI coworker. Give it a role and the tools that role uses; it does the work and keeps a record of everything.
저장소 README에서 가져옴
README 전체를 불러오는 중입니다.
README 전문 보기 ↗An assistant that pays attention — even when you're not asking.
One open-source assistant for your life and your work: run Jori as your personal assistant, or as your team's AI coworker. See getjori.ai.
Most assistants wait. You type, they reply, the moment passes. Jori stays on. It learns what you like, watches what matters, and quietly handles the things you'd rather not think about.
Runs on the Claude subscription you already have. No API keys. No per-token bills.
Formerly Edva, originally Merlin. The project was Merlin until 2026-10-25, Edva until 2026-10-28, and is now Jori, one name for the personal assistant and the business coworker. It lives at github.com/jori-ai/jori (old Edva and Merlin links point here). Existing installs: move the checkout to
~/Dev/jori, renameEDVA_*(orMERLIN_*) environment variables toJORI_*, rename theedva_*(ormerlin_*) Supabase tables tojori_*, and swap theai.edva.session(orai.merlin.session) launch agent forai.jori.sessionperdocs/installer.md.
→ Get started: docs/quickstart.md
You may have seen agent frameworks like openclaw. They're powerful, but they plug into the Claude API — separate account, pay per token, a bill that scales with how much your agent does.
Jori works the other way around. It runs through Anthropic's official Claude environment on the subscription you already use in your browser. Sign in once with claude, and Jori works through your account.
The agents run only on that subscription (a Claude Max login): Jori never quietly falls back to per-token billing, and if the subscription is unavailable it stops and tells you. An API key is optional, for a few developer tools such as jori review-pr.
Five things most AI assistants miss.
🧠 It remembers you. Your preferences, your decisions, the things you said last month. The longer you use it, the more it understands what you actually want.
🐝 It reaches out. When Jori notices something worth knowing, it tells you. Quietly. At the right moment. When there's nothing useful to say, it stays quiet.
🛠️ It gets things done. Doesn't just suggest. Writes the email, books the dinner, sets the reminder, runs the research. With safety rails for the things that matter — no money moves without you saying yes.
⏰ It shows up on time. Recurring work happens on schedule, in the background. You don't have to remember; it does.
📈 It gets better as you live with it. Every week, Jori reviews how it did, learns from how you responded, and tunes itself to fit your life more closely.
On a Mac, one line installs everything and opens a setup page in your browser:
curl -fsSL https://getjori.ai/install.sh | bash
Sign in with Claude, confirm a few details it already filled in, press Start Jori. Or brew trust jori-ai/tap && brew install jori-ai/tap/jori, npm create jori, or the Mac app from releases. Details: docs/quickstart.md. Later, jori update keeps it current.
Or step by step in the terminal:
# Clone this repository — copy its URL from the green "Code" button on GitHub
git clone <this repository's URL> ~/Dev/jori
cd ~/Dev/jori
# 1. Install deps, set up runtime dirs, check Ollama
./scripts/bootstrap.sh
# 2. Interactive setup — writes .env and agent/local/instance.json,
# generates a webhook token, seeds starter config from examples/
./bin/jori init
# 3. Check the install, then start everything in the foreground (Ctrl-C to stop)
./bin/jori doctor
./bin/jori up
# In another terminal — fire a test job
./bin/jori dispatch morning-digest
# Watch it work
./bin/jori tail ops
Cloned somewhere other than ~/Dev/jori? Export JORI_HOME=<path> (and keep it in .env) — the CLI, scripts and MCP configs resolve the install from it. For 24/7 deployment see docs/installer.md.
| Kernel | process-manager, supervisors, channel framework, MCP shell bridge, kernel test suite |
| CLIs | ~90 tools — jori (with Jori's own init, overlay, up, dispatch, tail, audit, review-pr, telemetry in bin/cli.d/), wiki, memory, email-*, context-*, hum-*, job-*, cron-*, self-improve-* |
| Libraries | instance (per-install identity + config), registry, wiki-store, local-llm, apns-push, supabase-rest, jori-context, the outbound/claim gates |
| Apps | Two SwiftUI companion apps — chat (apps/companion/) and fitness tracker (apps/trainer/) |
| Reference jobs | morning-digest, the hum family (hum, hum-review, hum-ideation, hum-daily, hum-consolidate, hum-explore), the weekly self-improvement loop (nightly-review, improvement-plan, system-issue-review, skill-repair, cost-efficiency-review), wiki upkeep, context-sync |
| CI | iOS + macOS TestFlight workflows, secret-scan, sanitize-guard, narrative review |
Jori is deliberately a small, opinionated kernel — not a turnkey product. The integrations are sparse because your assistant should know about your life.
agent/local/ (instance.json, shared.md, config/*.json), your own code in the _ext/ hook dirs. Both are gitignored; keep them in a private repo and link it with jori overlay add. See docs/architecture-public-private.md.docs/hum.md.docs/memory.md.jori plugin new <name>. See docs/writing-a-plugin.md (and docs/writing-a-job.md for playbooks).apps/companion/CODE-SIGNING.md.Minimum:
node:sqlite)gemma3:1b and nomic-embed-text pulledclaude)Full dependency matrix, per-feature requirements, and platform notes in docs/dependencies.md. The bootstrap script probes for what's installed and warns about gaps.
For the curious. Jori is built on Claude Code with persistent claude -p supervisors, durable session resume, an MCP shell bridge that turns any CLI into an agent tool, a SQLite memory layer with full-text + embedding search, a local Ollama model handling latency-sensitive acknowledgments, and markdown playbooks describing every scheduled job. Read the architecture: system/architecture.md.
Apache 2.0 — see LICENSE. Contributions welcome; see CONTRIBUTING.md. Security disclosures via SECURITY.md.