Your startup's agent team

Build with a startup team of agents.

A coordinated skill set for solo founders, developers, and startup teams—from strategy and product thinking through implementation and QA.

View on GitHub

Works across seven supported agent environments

One team, shaped around your role.

Choose where you sit. The same coordinated workflow changes what it emphasizes.

Move from idea to shipped product without pretending every specialist is in-house.

01

Pressure-test a product idea

02

Turn decisions into an approval-ready build brief

03

Ship with interface and QA checks

Every handoff stays attached to the same approved brief.

Startup TeamSimulated conversation · startup-team

Build an onboarding flow that gets a new user to first value in under two minutes.

I’ll turn this into one approved milestone plan, then start only the roles needed to ship it safely.

Approval requiredConfirm the brief and smallest safe role set before work begins.Review

I’m coordinating the work now.

Verified result

The final response will appear after the verification gates complete.

Clarifying your goal
Ask Startup Team a follow-up…

Example run · hardcoded preview

Use the agent environment you already trust.

Skills install across these agents. Host-managed internal role execution depends on the agent environment; public CLI dispatch is disabled.

YOUR GOALOMNISKILLSCOORDINATORPLANSPECIALIST SKILLBUILDSPECIALIST SKILLVERIFYSPECIALIST SKILLVERIFIEDRESULT

Everything your startup team needs.

Specialized skills, one shared context, and visible quality gates from first decision to release.

Direction to delivery6 coordinated capabilities

Strategy & validation

Challenge assumptions and make a defensible direction decision.

Product requirements

Produce clear scope, acceptance criteria, and approval-ready requirements.

Interface design

Shape responsive hierarchy, interaction, accessibility, and motion.

Architecture & implementation

Expose technical risk and execute the approved build slice.

QA & release verification

Check acceptance, regressions, responsiveness, and release behavior.

Approval gates & handoffs

Keep decisions and accountable roles visible throughout the run.

A system for shipping with agents, not just chatting with them.

Every specialist works from the same decision trail, with your approval where it matters.

Feature 01

Start with the outcome, not a maze of prompts.

Startup Goal clarifies the request once, keeps the decision context together, and routes only the specialists the work needs.

Omniskills Teams

Pick the team for the goal

Install a real coordinator and its specialist roles together. Startup Team ships product work; Finance Team researches public companies; Market Team builds a sourced regime view.

Featured team

Startup Team

Move one approved startup feature at a time through plan approval, implementation, conditional rework, QA, User Outcome Replay, and human feature acceptance.

Coordinator
$startup-goal

Controls the Goal Tunnel and Evidence Ledger, runs selected installed roles as internal subagents by default, and holds both human approval gates. A run-scoped visible_task needs explicit post-plan confirmation and uses gpt-5.6-terra with reasoning effort high in a project/worktree-aware workspace-write boundary. Prepared, not executed names unavailable capability, exact-model, target/access, creation failure, reconciliation, or result collection; public CLI dispatch stays disabled.

Members

  • CEOCompany direction and tradeoffs
  • CTOArchitecture and technical risk
  • Product ManagerDiscovery, PRDs, and issue slicing
  • Web DesignInterface direction and motion quality
  • Engineering ManagerDelivery sequencing and quality gates
  • Founding EngineerImplementation framing and handoff
  • Deliver CodeApproved vertical slices through TDD, review, and verification
  • QA LeadAcceptance checks and release risk

Research team

Finance Team

Local preview: prepare manual company, financial, valuation, and risk handoffs, then combine completed outputs into one sourced public-company brief.

Coordinator

$finance-research

Company Analysis · Financial Analysis · Valuation Analysis · Risk Analysis

Research team

Market Team

Local preview: prepare manual macro, rates, structure, sector, and risk handoffs, then combine completed outputs into one sourced regime brief.

Coordinator

$market-research

Macro Analysis · Rates Analysis · Market Structure · Sector Analysis · Risk Analysis

Skill Hub

Explore the Skill Hub

Browse independently installable workflows or inspect the skills they assemble.

8 workflow results

Install once. Invoke one goal. Stay in control.

Three deliberate steps keep the work understandable from setup to verified release.

  1. 01

    Install the team

    Copy one command into a supported agent environment.

  2. 02

    Call $startup-goal

    Describe the outcome; the coordinator clarifies until the brief is ready.

  3. 03

    Approve, verify, and accept

    Approve the implementation boundary, then review QA evidence and User Outcome Replay before feature acceptance.

Common commands

Get up and running fast

Install by alias, public git URL, or local path. The CLI validates the workflow manifest, bootstraps missing external skills from workflow metadata, and records installed Omniskills workflows under ~/.omniskills/workflows/ by default. Loop-enabled workflows use omniskill loop for resumable, action-only state.

Then invoke in your agent

>/startup-goal I have an AI bookkeeping idea; help me choose the wedge and ship a v1 in two weeks
 
[ok] Intake clarified customer, deadline, and non-goals
[ok] CEO chose a narrow first wedge
[ok] PM scoped receipt capture to month-end summary
[ok] CTO kept finance automation behind seams
[ok] EM sequenced prototype, pilots, and release gate
[ok] engineer selected the receipt-to-summary slice
[ok] QA checked privacy, totals, and recovery paths

Author your own workflow

Package your workflow as an Omniskills workflow

Scaffold a bundle, define the entry skill, list sub-skills in workflow.json, validate, and share.

Author guide

Clear answers before install.

What changes, what stays under your control, and what the team actually does.

Does Startup Team replace my judgment?

No. You approve the milestone plan and feature acceptance. Any material scope change returns to planning.

How are Startup Team installs and local research previews kept safe?

Startup Team carries a checked-in schema 0.2 lock with exact-commit external locators, and its declared members resolve from the same checkout. Managed refreshes require recorded ownership; mixed ownership fails closed. Finance Team and Market Team remain lockless local previews.

Which CLI command installs a workflow or team?

Use install as the public install command. bundle and workflow remain compatibility aliases.

Which agents are supported, and which features differ by agent?

Skills install into Cursor, Codex, Claude, OpenCode, Hermes, OpenClaw, and GitHub Copilot. Host-managed internal role execution requires an agent environment with internal subagent support; public CLI dispatch is disabled.

Can I inspect every included skill?

Yes. Startup Team exposes its coordinator, member roles, workflow source, and local skill sources for inspection.

Does every goal run every specialist role?

No. The coordinator selects the smallest safe role set and explains which roles were skipped and what would bring them back.

Give your next goal a real startup team.

Install the team once, call one goal, and keep every decision and verification step visible.

View source