Visit - https://engine-user.github.io/Prometheus-AI-Agent/ - for more details
Meet Prometheus — Your own a local-first personal assistant that shows the four pillars behind every serious agent: Harness · Loop · Memory · Eval/LLM-Ops. No frameworks hiding the good parts.
- Local-first. Your memory is one SQLite file. Open it. Read it. It's yours.
- Memory is the hero. Semantic + episodic + procedural — with a gate that decides whether to remember, and a pass that decides what to keep.
- The loop is ~95 lines of plain Python. Step through it.
- Watch it think. A local dashboard lights up every message as it flows through the harness.
- Eval built in. Deterministic tests and LLM-as-judge, side by side, with a release gate.
Just want to run it:
pip install prometheus
prometheus # talk to your Prometheus in the terminal
prometheus dashboard # …or the browser cockpit → localhost:7777It will tell you which key to set the first time. Want to read the code (the point of this repo) or contribute — clone it instead:
git clone https://github.com/Engine-User/Prometheus-AI-Agent && cd Prometheus-AI-Agent
uv venv && uv pip install -e . # create the env + install the `prometheus` command
cp .env.example .env # pick a provider, paste ONE key
uv run prometheus # talk to your Prometheus in the terminal
uv run prometheus dashboard # …or the browser cockpit → localhost:7777Now try it. "Remember that Engineer prefers morning meetings." Quit. Restart.
"Book a catch-up with Engineer on Friday." → it remembers, and books 9am. Your memory is one
file: ~/.prometheus/state.db, the same from every folder.
Use the model you already pay for. Anthropic (default), OpenAI, Gemini, DeepSeek, MiniMax,
Kimi, GLM, OpenRouter (one key, hundreds of hosted models), OpenCode Zen, or OpenCode Go—
set PROMETHEUS_PROVIDER=, paste the key, done. One dialect in the loop;
a ~60-line adapter handles the rest.
New to it? Getting started walks the whole setup, with a check at the end of every step.
Prometheus's own memory is local. But you can connect to other sources of memory as you like, locally.
pip install 'prometheus[mcp]' # in a checkout: uv pip install -e '.[mcp]'
prometheus connect prometheus-memory # or any other memory store.
prometheus skill export --to claude,codex # carry Prometheus's skills to Claude Code and Codex too| Pillar | In one line | Read more |
|---|---|---|
| Harness | gateways (terminal, dashboard, voice, Telegram, Discord, WhatsApp) and tools around one loop | architecture |
| Loop | ~95 lines of plain Python: reason, act, repeat, with two ways to stop | the tour |
| Memory | semantic, episodic and procedural (skills); a gate decideswhether to remember, consolidation decides what to keep | the tour |
| Eval / LLM-Ops | deterministic tests and LLM-as-judge side by side, a release gate, a trace for every turn | evals |
How is this different from ChatGPT or Claude Desktop? Those are products you use. This is a codebase you own: the loop, the memory schema, the gate and the eval harness are all yours to read and change. Versus the big open-source assistants (OpenClaw, Hermes)? Same architecture, 1/100th the code.
| Read | For |
|---|---|
| Getting started | installing, the first run, connecting Prometheus Memory |
| The tour | the dashboard, things to try, the loop, graph workflows, skills |
| Architecture | every box on the whiteboard, and the file behind it |
| Integrations | voice, Telegram, calendars, MCP servers, Prometheus Memory |
| Commands | everyprometheus and make command |
| Evals & tracing | the two kinds of eval, the Docker tier, the release gate, traces and spend |
| Roadmap | what is live, what is still a skeleton, upgrade paths |
| Whiteboards | the editable system-design charts from the videos |
| lab/ | Prometheus meets other agents and models: the video experiments |
| AGENTS.md | the rules, and how to send a PR |