Orchestrate Complex Objectives into Guided Sequences
Meta-orchestrator that turns a complex objective into a guided, step-by-step sequence of other skill invocations.
Why it matters
Automate multi-phase objectives by orchestrating a sequence of skill invocations. This skill guides users through complex tasks like shipping an MVP, conducting security audits, or building AI agent systems.
Outcomes
What it gets done
Combine multiple skills into a single, guided execution flow.
Execute best-practice workflows for common scenarios (SaaS MVP, security audits, AI agents, browser QA).
Automate browser interactions and end-to-end quality assurance.
Provide structured output including artifacts, validation evidence, and risk assessments.
Install
Add it to your toolbox
Run in your project directory:
curl -fsSL https://spark.entire.vc/get/ag-antigravity-workflows | bash Overview
Antigravity Workflows
A meta-orchestrator skill that turns a complex, multi-phase objective into a guided, step-verified sequence of other skill invocations. Use when combining multiple skills for a multi-phase goal; it orchestrates rather than replaces specialized skills, which must be locally available.
What it does
Antigravity Workflows turns a complex, multi-phase objective into a guided sequence of other skill invocations, so the user doesn't have to manually pick each skill themselves. It reads workflow definitions from docs/WORKFLOWS.md (human-readable playbooks) and data/workflows.json (machine-readable metadata) as its source of truth, then: identifies the user's concrete desired outcome, proposes the 1-2 best-matching workflows, has the user choose one, and executes it step by step - announcing the current step and expected artifact, invoking the recommended skills for that step, and verifying completion criteria before moving to the next. At the end it reports completed artifacts, validation evidence, and remaining risks/next actions.
Default workflow routing covers common scenarios: a product delivery request routes to "ship-saas-mvp", a security review request to "security-audit-web-app", an agent/LLM product request to "build-ai-agent-system", an E2E/browser testing request to "qa-browser-automation", and a domain-driven design request to "design-ddd-core-domain".
When to use - and when NOT to
Use this when the user wants to combine several skills without manually selecting each one, or when the goal is multi-phase (plan, build, test, ship) - for scenarios like shipping a SaaS MVP, running a web security audit, building an AI agent system, or implementing browser automation and E2E QA.
This skill orchestrates rather than replaces specialized skills - it depends on those referenced skills being locally available, doesn't guarantee success without the required environment access, credentials, or infrastructure, and for stack-specific work (e.g. Go browser automation) may need a corresponding skill like go-playwright present in the local skills repository.
Inputs and outputs
Inputs: a user's concrete multi-phase objective (e.g. "ship a SaaS MVP," "run a security audit").
Outputs: a chosen workflow executed step by step, with completed artifacts, validation evidence per step, and a final summary of remaining risks and next actions.
Integrations
Depends on the local availability of the specialized skills each workflow invokes (e.g. skills for SaaS MVP delivery, web security audits, AI agent building, browser automation/QA, or DDD core domain design); related planning skills include concise-planning, brainstorming, workflow-automation, and verification-before-completion.
Who it's for
Users working across a local skills repository who want a guided, step-verified path through a multi-phase objective rather than manually sequencing individual skills themselves.
FAQ
Common questions
Discussion
Questions & comments · 0
Sign In Sign in to leave a comment.