One platform for the work that actually matters.
OpenLotus pairs with your real environment, reads the work you actually do, and turns it into a living plan. Progress map, dashboard, and agent workspace in one loop.
A drop-in skill that turns your agent into a graph-engineered builder. Paired with OpenLotus memory.
Works with Claude Code, OpenCode, Cursor. Zero repo permissions.
Works with your agent environment today
No plugins. The skill drops into any terminal agent; the MCP server talks to anything that speaks MCP. Your local git is the first reality source.
Connectors on the roadmap
The whole loop, in one place.
Research, plan, build, learn. OpenLotus turns the work you already do into a plan you can trust.
Three surfaces, one ground truth
Progress map, dashboard, and agent workspace all read the same record of your real work, so nothing gets re-explained.
It watches what you build
Files you touch, decisions you make, work you haven't committed yet. The map fills itself from your actual week.
Momentum you can see
See the week form as it happens. Know what moved, what stalled, and what's next.
Your work stays yours
No repo permissions, no scanning, no exports. OpenLotus lives next to your editor, not in your way.
How OpenLotus works.
PairOne CLI command connects OpenLotus to the codebase you're building in. No repo permissions, no scanning.
Watch Files you touch, TODOs you leave, decisions you make. The progress map fills itself from your real week.
ActAsk your agent what's next. It reads the same ground truth you do. No guessing, no re-explaining.
One loop.
Three surfaces. One ground truth.
Progress Map
Stop relying on your agent to remember everything. Before, your agent held the context and when it lost track, you did too. Now you both see one living tree: what you planned, what you decided, and what actually changed. When the agent drifts, you catch it here.
OpenLotus doesn't eat your context window. A memory layer holds your workspace conversations and pulls in live signals from your code through MCP. Then it turns all of that into the tree you see below.
- You and your agent see the same plans, decisions, and changes
- Drift shows up when the agent heads in the wrong direction
- Your context lives in the map, not just the chat window
Dashboard
Your week, summarized.What changed, what's drifting, and what to do next. Written for a founder, not a project manager.
- Weekly summary written for you, not a project manager
- What changed, what drifted, what to do next
- Monday readout, ready when you are
Agent Workspace
Your agent, in context. The agent reads the same ground truth you do, so nothing gets re-explained.
- Your agent reads the same ground truth you do
- Tool calls you can trace back to their source
- Follow-up suggestions based on what just happened
Connectors
Your stack, connected. Pair in one command and keep the loop live with MCP. OpenLotus never needs repo permissions.
- Connect your stack in one command
- MCP keeps the loop live
- No repo permissions needed — ever
vibe-to-ship.
A drop-in skill that turns your agent into a graph-engineered builder. It breaks work into clear tasks, runs them in parallel only when they must wait on each other, and checks its own work with a fresh pair of eyes.
On its own it’s powerful. Paired with OpenLotus memory, every plan it makes is anchored in your real repo and every result lands where you can see it — a system you can’t get from a plain chat.
Install the skill, then say “set up OpenLotus” and it wires the rest. Or do the three copy-pastes by hand — same result, one time. See both paths.
Free. Drop-in skill. Works anywhere.
Wire your agent in.
Three steps, one time.
OpenLotus lives in your agent, not in another browser tab. Paste two small files, run one command, and every session from now on reads and writes shared memory automatically.
- 1Pair this repoTerminal (once per repo)
npx openlotus pair
From
openlotus-frontend: opens your browser to pick a project. No flags needed. From repo root: addopenlotus-frontend/prefix. Afterpnpm link, justopenlotus pair. - 2Register the MCP servermcp.json
{ "mcpServers": { "openlotus": { "command": "node", "args": ["/absolute/path/to/openlotus/cli/mcp.mjs"], "env": { "OPENLOTUS_BASE_URL": "https://www.openlotus.io" } } } }Makes get_memory / get_reality / get_drift / record_decision available in every session.
- 3Give your agent the rulesCLAUDE.md / AGENTS.md
## OpenLotus loop (standing rules) Shared memory lives in OpenLotus via MCP tools: get_memory, get_reality, get_drift, record_decision. The founder sees the same data on the web dashboard. 1. Boot: at the START of any work session, call get_memory + get_drift before planning anything. Surface drift findings to me first. 2. Decisions: whenever a direction choice is made (stack, scope, schema, tradeoff), call record_decision immediately — title + why. 3. Reality check: after meaningful work, call get_reality; mention surprises. 4. PROGRESS.md: keep it in sync with the map after each milestone. 5. Stale > silent: if context is missing, call get_memory again — never guess.
One-time paste so every session boots knowing the loop — never re-paste into prompts.
Your stack, already in the loop.
Pair once and OpenLotus watches the tools you already use. No reconfiguration, no exports.
See connectorsThe old way vs. the loop.
More tools meant more context to lose. OpenLotus keeps everything in one running record fed by your real work.
Scattered tools
The usual way
OpenLotus
One loop
Ready to build with momentum?
Pair your environment in one command and let OpenLotus turn your real work into a plan.
Built for the loop, not the meeting.
Everything you need to move research, plan, build, and learn forward without leaving your editor.
Pair in one command
CLI pairing, MCP live watch, and connectors that never ask for repo access.
Live, from your editor
The loop updates as you work. No exports, no sync buttons, no busywork.
CLI-first
One command to pair. MCP to keep it live. Docs and API when you need them.
Private by default
OpenLotus reads work signals, never your code content. Zero repo permissions.
Built around your real week.
OpenLotus turns the work you already do into a plan you can trust. No extra data entry, no busywork.
1 command to pair your environment.
3 surfaces, one shared loop.
0 repo permissions, ever.
What early builders say
Founders who paired their environment and let the loop do the talking.
Solo founder Building an AI tool
The progress map fills itself from the work I actually do. No more status updates.
Indie hacker Shipping weekly
My agent stopped asking me to re-explain my week. It watches the loop and already knows.
Platform builder Developer tools
One command to pair, and the loop just started building itself. I keep checking it works.
Product lead Small team
I open the dashboard on Monday and it tells me what changed. That is the whole job done.
Simple, honest pricing.
Start free, upgrade when you need more power. No connector fees. Connectors are free forever.
Free
For founders just starting. No credit card required.
- 1 product
- 2 connectors
- OpenLotus agent
- 500 memory events
- Auto-summarized forever
- Connector request form
Pro
For growing products that need the full picture.
- 3 products
- Unlimited connectors
- All 5 agentsSoon
- TimeFM forecastingSoon
- 5k memory events
- Auto-summarized forever
- Ecosystem playbooksSoon
- BenchmarkingSoon
Studio
For agencies managing multiple client products.
- Unlimited products
- Custom agentsSoon
- 3 team seats
- White-labelSoon
- Priority support
- Unlimited memory
FAQs
Your questions answered
Need more help? Contact us
Build with momentum.
Pair your environment in one command and let OpenLotus turn your real work into a plan.
Get started free