Start with AGENTS.md at the repo root: it routes every kind of change to the one file that covers it. This folder holds those files, sorted into four groups.
- getting-started.md — install, the first run, and connecting Prometheus Memory, with a check at the end of every step.
- tour.md — the dashboard, things to try, the loop, graph workflows and skills.
- status.md — what works, what is known-broken, what is deliberately not built. Rewritten whole rather than appended to, so it cannot become a changelog. Read it before opening a PR.
| file | what it answers |
|---|---|
| context/conventions.md | how much process a change needs, where capability goes, testing, git, scope |
| context/design-system.md | how the dashboard looks, and which primitive to use |
| context/writing-rules.md | how we write docs, UI copy and commit messages |
| context/gotchas.md | traps someone already stepped on |
| context/maintainers.md | how maintainers review, merge and release; contributors can skip it |
| file | what it answers |
|---|---|
| architecture.md | the four pillars, and which file is which diagram box |
| loop-vs-graph.md | when a turn needs shape, and why the loop never changes |
| agent-graphs-design.md | the graph engine's design and its fail-open rule |
| providers-registry.md | adding a model provider: one table inprometheus/providers.toml |
| memory-backends-playbook.md | seeing your memories in each provider's own console |
| benchmarks.md | what has been measured, and how |
| integrations.md | voice, Telegram, Apple, Google Calendar, MCP, Prometheus Memory — all opt-in |
| commands.md | everyprometheus and make command |
| evals.md | the two kinds of eval, the Docker tier, the release gate, traces and spend |
| roadmap.md | what is live, what is still a skeleton, upgrade paths |
Every whiteboard is an editable .excalidraw source: download one, drop it
on excalidraw.com, and remix it for your own team.
The ones that explain Prometheus live in whiteboards/; the ones from
videos about other projects live with their topic in ../lab/.
| Chart | What it explains |
|---|---|
prometheus-architecture.excalidraw |
Prometheus itself — harness, loop, memory pillars, LLM Ops (editable rebuild ofthe whiteboard); its LLM Ops panel is the whiteboard's, with observe, two evals and a release gate, not the dashboard's four boxes (Trace, Observability, Evals, Release) |
loop-vs-graph.excalidraw |
Loop vs graph engineering — the ladder, and two timelines from a measured run ofprometheus brief against prometheus gather (the write-up) |
k3-architecture.excalidraw |
Kimi K3: the 16-of-896 MoE, KDA + AttnRes attention, why agent loops get cheap |
pi-architecture.excalidraw |
pi (72K-star coding agent): 4-tool core, extensions, one EventStream |
New charts arrive with every video. If they help you, a star keeps them coming — and sponsoring gets new whiteboards early.
scripts/whiteboard/build_*.py generates the Prometheus boards so they match the
hand-drawn masters; edit the builder, not the JSON.