Every conversation starts from zero
Strategy, pipeline, and clients live in different notes. AI answers your latest prompt, not the state of your business.
The founder operating system
Founder OS turns your goals, cash, pipeline and commitments into one daily decision — stored locally and traceable to its source.
Local Markdown · No automatic sending · Explicit ownership · No hidden actions
08:00 / Daily brief
Thursday / Pipeline review
16:00 / Weekly review
The real problem
You need a system that remembers what you decided, spots contradictions, and brings the decision back when it is actually due.
Strategy, pipeline, and clients live in different notes. AI answers your latest prompt, not the state of your business.
The brief sounds smart, but has no owner, date, or next move. A week later, nobody comes back to it.
You set the priority, chase the sale, calculate the margin, and judge your own plan. That is not a lack of ambition. It is a lack of structure.
What makes it different
Founder OS does not pretend to be a magic autopilot. It builds continuity: shared state, clear ownership, operating cadence, and hard boundaries.
Goals, metrics, pipeline, clients, decisions, and drafts live in a local Markdown workspace. Tomorrow starts with business state, not another retelling.
The CFO does not rewrite the pipeline, and the Brand Editor does not change strategy. A write-time guard keeps every role in its lane.
A brief in the morning, pipeline review on Thursday, weekly review on Friday, and financials after month-end. Run it by hand or on a local schedule.
Every role has principles you may disagree with. A plan gets a verdict, a bet gets a kill condition, and “no time” gets a calendar audit.
A real workspace, with fictional data
Studio North is a one-person strategy studio. Its example workspace is small enough to inspect, but complete enough to show how a bet becomes a commitment, a calendar block, and the next morning's brief.
The one thing
q-0720a Finish the Acme proposal scope — serves B1.
Rotting
Northwind reference follow-up — 3 days overdue.
The trade
The website redesign does not start. It serves neither quarterly bet.
Doing
q-0720a Finish the Acme proposal scope — bet: B1 — started 2026-07-20.
Queued
q-0717b Publish the Northwind migration case study — bet: B2.q-0717c Book two qualified discovery calls — bet: B1.Dropped
Website redesign — passed over by three weekly reviews and serves no current bet.
Threshold · 30 September
Sign 3 sprints at or above $15,000 without adding custom delivery.
Downside cap
40 founder hours and $2,000 in design support.
Kill condition · 31 August
Fewer than 2 qualified proposals.
Arithmetic
40h available · 24h delivery · 16h free · 9h planned.
The plan uses 56% of free capacity instead of pretending every free hour is deliverable.
Selected funded blocks
| Day | Block | Serves |
|---|---|---|
| Mon · 09:00–11:30 | Finish Acme proposal scope | B1 |
| Tue · 09:00–11:00 | Interview case-study client | B2 |
| Thu · 10:00–11:00 | Pipeline review | B1 |
Pipeline · $33,000 live
| Deal | Stage | Next action |
|---|---|---|
| Acme · $18k | Proposal | Finish scope · Jul 20 |
| Northwind · $15k | Qualified | References · Jul 17 · overdue |
| Delta · $12k | Dead | Open-ended terms broke the offer boundary |
What was learned
Delta was lost because the buyer required ongoing availability, not because the price was too high.
June close
| Booked | $31,000 | Collected | $24,000 |
|---|---|---|---|
| Hours | 112 | Effective rate | $214/h vs $200 target |
| Receivables | $7,000 | Runway | 9.7 months |
Profitability signal
The fixed-scope sprint is above target. Custom advisory added 14 unbilled hours and fell below it.
Context
The fixed-scope sprint cleared the $200/h target. Custom advisory fell below it after 14 unbilled hours.
Rejected
Keep a $10,000 entry sprint and upsell advisory later.
What would change our mind
Fewer than 2 qualified proposals by 31 August, provided at least 8 ICP-matched discovery calls happened.
How to start
Install the plugin, run one resumable onboarding flow, and start from a persisted daily decision. It targets a ten-minute median and stops at fifteen minutes rather than expanding into setup theatre.
Add the marketplace, then install the plugin. No Founder OS account, dashboard, or external database.
/plugin marketplace add msolecki/founder-os/plugin install founder-os@founder-osBusiness, customer, quarter, and money are routed to their file owners. Unknowns stay unknown. A valid reviews/daily/YYYY-MM-DD.md has all four required headings: ## The one thing, ## Rotting, ## The trade, and ## Triage, with non-empty ## The one thing and ## The trade.
/founder-os-initOptional cron cadences run on your machine while it is on. Every workflow also works manually without cron.
/setup-cadencesThe business in motion
Founder OS does not create more work. It connects the moments when you already need to make a decision.
The Chief of Staff reads your goals, metrics, and weekly plan, then names the most important move.
/daily-briefThe Focus Coach turns quarterly bets into concrete blocks of work.
/week-planThe publishing plan starts with real audience questions, not an empty content slot.
/content-planThe Pipeline Coach forces a next move. Conversations without action leave the pipeline.
/pipeline-reviewCommitments against outcomes. No copying everything into next week by default.
/weekly-reviewThe complete operating library
Start with the decision in front of you. Founder OS routes it to the right owner and workflow.
/daily-briefscheduledOpen the day with the one move that matters most, grounded in current goals, metrics, and this week's plan.
/weekly-reviewscheduledScore what you committed to against what actually happened, then name the pattern that keeps repeating.
/monthly-reviewRead the month back against the company charter and identify drift after the financial close is complete.
/decision-logRecord an irreversible decision, why it was made, and the evidence that would justify reversing it.
/triageTurn a pile of obligations into one priority, price the cost of the rest, and route each item to an owner.
/queueKeep live obligations capped, dated, owned, and able to expire instead of becoming an endless wish list.
/quarterly-planningscheduledClose the old quarter's bets and commit to at most three new ones, each with a measurable kill condition.
/bet-sizingPrice what a bet costs when it is wrong and cap the downside before it enters company goals.
/kill-or-continueForce a verdict when the judgment date arrives, using the original threshold and naming sunk cost out loud.
/annual-reviewReview twelve months of decisions and score the quality of judgment, not whether luck produced a good outcome.
/red-teamAttack a finished plan as a hostile reader and return a verdict, not a polite list of concerns.
/assumption-auditList what must be true, remove what already has evidence, and rank the rest by the cost to test.
/premortemAssume the plan is dead six months from now and tell the failure story while it can still change the plan.
/icp-definitionNarrow the ideal customer until the definition excludes real, nameable people—not just vague adjectives.
/offer-designTurn bespoke work into a clear buyer outcome with an explicit boundary a prospect can repeat.
/pricing-strategyPrice against the buyer's outcome and real alternative, then put your walk-away floor in writing.
/pipeline-reviewscheduledGive every deal a next action with a date—or remove it from the pipeline.
/outreach-draftDraft a first contact or follow-up from the prospect's problem, never from “just checking in.”
/proposal-draftDraft a qualified proposal with scope, price, exclusions, and an expiry date after capacity is confirmed.
/win-loss-analysisReconstruct why a deal ended from the record and the prospect's own words while the evidence is fresh.
/capacity-checkCompute real deliverable hours before saying yes to a client, extra scope, or a start date.
/scope-guardRule whether a client request is inside scope by checking it against the proposal and its exclusions.
/client-healthScore payment, scope, tone, and effort before an engagement turns into a crisis or reaches renewal.
/delivery-retroCompare estimated with actual hours within five days of shipping, before memory rewrites the project.
/revenue-reviewscheduledClose the month on revenue booked, cash collected, and the effective hourly rate actually earned.
/runway-forecastCalculate months of survival at real burn, discounting pipeline value by the stage of each deal.
/profitability-analysisRank clients by effective hourly rate to show exactly where margin is disappearing.
/rate-raiseDecide whether a rate should rise, by how much, and prepare the script when the numbers justify it.
/week-planscheduledTurn this quarter's bets into dated blocks before the week fills itself with other people's priorities.
/calendar-auditscheduledCompare where the week was meant to go with where it actually went, using the calendar as evidence.
/energy-auditFind the hours when your output is consistently good or poor from the accumulated calendar record.
/skill-gapName the capability gap between the offer sold today and the offer current bets require.
/learning-planAttach one needed capability to one real project and one deadline that forces practice.
/content-planscheduledPlan what gets published against the ICP and the number of pieces you actually shipped last month.
/voice-captureLearn your voice from writing you really sent, including the edits you made before sending it.
/content-draftDraft one planned piece around one useful idea you learned by doing—not to fill an empty slot.
/audience-researchCollect the exact language your ICP uses for the problem before deciding what to publish.
/relationship-mapMap real relationships by category and days since meaningful contact before reaching for cold outreach.
/follow-up-sweepscheduledSurface up to five overdue relationships and give each person a real reason to hear from you.
/automation-auditUse payback arithmetic to decide whether a manual task is worth automating before building anything.
/tool-stack-reviewFind software you pay for but do not use and prepare a cancel list before renewals arrive.
/portfolio-reviewscheduledRank active businesses, split this week's hours and cash, and say plainly what that allocation is starving.
/context-loadLoad dated charter, goals, and metrics before advice begins, including the active business slug when needed.
/founder-os-doctorDiagnose workspace rot—missing files, stale metrics, orphan clients, broken bets, and silent cadences—before repairing it.
/founder-os-initInterview the founder, scaffold the workspace, route every answer to its owner, and produce the first brief.
/guardrailsEnforce the hard refusals: nothing outbound, nothing paid, and professional escalation for regulated advice.
/ingestion-gateClassify every external claim as fact, validate, or disregard before it enters canonical company state.
/setup-cadencesSchedule nine recurring reviews per business, plus one portfolio review when multiple active businesses make it relevant.
/state-integrityResolve every workspace write against the ownership map and hand it to the correct role when needed.
No workflows match that search and filter.
Commands are shown in their short form. The namespaced form is /founder-os:<workflow>. Open the generated command reference →
Your executive team
Every agent owns exactly one decision nobody else can make. When you do not know where to start, the Chief of Staff routes the question to the right role.
1role = one decision = one owner of state
Built for a portfolio of one
Founder OS repeats the operating system instead of blending your companies together. Every business keeps its own goals, clients, metrics, pipeline, decisions, and cadence.
Run one business? Nothing changes. The multi-business layer appears only when you register a second active business.
businesses.yaml
one full workspace per business
Its own operating state and its own scheduled rhythm.
A separate history. No shared prospects, numbers, or obligations.
one deliberately narrow cross-business view
Target any workflow: /founder-os:daily-brief studio
One source of truth
It lives in simple files you can open, edit, archive, and version. Every material claim carries a source and date; agents read the same state, but write only inside their own area.
Founder OS only knows what is recorded here. It does not automatically sync your calendar, CRM, inbox, or bank account.
Workspace files stay on your machine. Prompts and context sent through Claude Code or Codex remain governed by that environment's data-handling terms.
Hard boundaries
No agent has a browser, shell, or tool that can reach beyond the workspace. That is a deliberate boundary, not a missing feature.
The system can prepare a proposal, follow-up, or rate-raise plan. You press the button, sign the document, and own the decision.
An honest filter
Before you install
Founder OS is free and MIT-licensed. It runs inside your existing Claude Code environment and adds no account or subscription of its own.
Continue to installationFounder OS is a plugin, not a standalone app. Open Claude Code docs.
Python runs the ownership hook. PyYAML enables its full ownership-map check.
Manual workflows need no cron. Scheduled cadences run only while your machine and cron service are on.
Founder OS is free. Your existing Claude Code plan and usage remain separate.
FAQ
No. It is a Claude Code plugin. The agents, workflows, and rules install inside that environment, while your business state stays in a local Markdown workspace.
No. They are specialized roles, not 13 autonomous processes running all day. A workflow invokes the relevant role; nine per-business cadences can run on a local schedule, with Portfolio Review as the tenth on multi-business installs.
The plugin installs with two commands after adding its marketplace. Onboarding targets a ten-minute median and stops at fifteen minutes. It is complete only when the first valid daily brief is persisted and passes the same validity check.
No. When you do not know what to use, ask the Chief of Staff. Routing is its job — it sends the question to the right agent.
Yes. The technical section headings stay in English, but the workspace content and the agents' work can use your language. The system can also learn your style from real writing samples.
Each business gets its own workspace and cadences. The Portfolio Manager is the only role that looks across businesses, dividing time and cash without mixing operational state.
Only from the Markdown workspace and context you provide. It does not automatically sync your calendar, CRM, inbox, or bank account. Workspace files stay on your machine; your Claude Code or Codex environment's data-handling terms still apply to prompts and context sent through it.
Founder OS is free and MIT-licensed. It adds no account or subscription. Your existing Claude Code plan and usage are separate.
After the optional /setup-cadences, local cron runs nine cadences per business while your machine and cron service are on. Multi-business installs add Portfolio Review as the tenth. There is no cloud scheduler, and every cadence also works by hand.
Start with one decision, not another dashboard
/plugin marketplace add msolecki/founder-os/plugin install founder-os@founder-os/founder-os-initThen: onboarding runs one continuous, resumable path and ends only after your first valid daily brief is persisted and validated. Read the complete getting-started guide for the data boundary, first five actions, update, repair, and uninstall steps.