Gaia

full-stack apps. enterprise-grade. maintainable. customizable.

A team of AI agents, available as a plugin for GitHub Copilot and Claude Code.

---

---

What is Gaia?

Gaia is a team of AI agents that builds and evolves software using spec-driven development. You describe your goal; Gaia coordinates intake, architecture, planning, implementation, testing, and release — enforcing QA and release gates along the way.

The workflow contract lives in AGENTS.md. Gaia ships as three composable plugins, backed by a custom .NET MCP server (src/) that persists tasks, memory, and self-evolution lessons across machines and sessions.

---

Catalog

Plugins

Plugin	When to use
foundation	Always — base layer for any solution. Authoring new agents/skills/plans and wiring the remote MCP server.
engineering	Building or evolving software end-to-end: intake → architecture → planning → implementation → testing → release.
product	Validating a consumer one-off / in-app-purchase product idea against real net-of-fee economics before you build.

Engineering agents

Agent	When to use
fa-intake-coordinator	Entry point — frames a raw request, clarifies scope, and routes work to the right specialists.
fa-solutions-architect	Designing the system: drafting/updating architecture specs in docs/ before code.
fa-implementation-planner	Breaking approved architecture into a sequenced, gated task plan.
fa-software-engineer	Implementing planned tasks and writing the code.
fa-tester	Writing tests and validating behavior against QA gates.
fa-release-engineer	Verifying release gates and shipping the change.

Product agents

Agent	When to use
fa-product-coordinator	Entry point — frames the brief and orchestrates the 10 money-gated discovery stages.
fa-product-ideation-strategist	Generating and shaping product concepts.
fa-product-market-researcher	Sizing demand, competition, and market opportunity.
fa-product-validation-researcher	Testing demand signals and de-risking assumptions with evidence.
fa-product-monetization-economist	Designing pricing and the unit-economics model.
fa-product-build-architect	Scoping the vertical slice / MVP and production hardening.
fa-product-liveops-manager	Planning post-launch live-ops and retention.
fa-product-growth-marketer	Designing go-to-market, soft launch, and acquisition.
fa-product-compliance-officer	Checking platform, legal, and store-policy compliance.
fa-product-reviewer	Adversarially reviewing each stage's output.
fa-product-synthesizer	Reconciling fan-out + review into a single decision against the economics model.

Skills

Skill	Plugin	When to use
fa-create-agent	foundation	Authoring or revising a custom agent definition.
fa-create-skill	foundation	Authoring or revising a SKILL.md.
fa-create-plan	foundation	Researching and outlining a multi-step plan.
fa-process	engineering	Orchestrating the end-to-end software workflow and gates.
fa-architecture	engineering	Keeping architecture docs aligned before structural changes.
fa-default-tech-stack	engineering	Choosing the baseline stack when none is specified.
fa-planning	engineering	Turning architecture into a branch-aware, gated task plan.
fa-engineering	engineering	Delivering planned implementation against repo conventions.
fa-ui-engineering	engineering	Building React UI with the design system and semantic tokens.
fa-testing	engineering	Running formal validation layers and recording proof.
fa-product-process	product	Orchestrating the money-gated product lifecycle.
fa-product-discovery	product	Researching market and opportunity.
fa-product-ideation	product	Generating and shaping concepts.
fa-product-monetization-design	product	Designing pricing and monetization.
fa-product-validation	product	Validating demand with evidence.
fa-product-vertical-slice	product	Scoping the MVP / vertical slice.
fa-product-production-hardening	product	Hardening the build for production.
fa-product-compliance	product	Checking compliance and store policy.
fa-product-soft-launch	product	Planning and running a soft launch.
fa-product-gtm-launch	product	Planning the full go-to-market launch.
fa-product-liveops	product	Planning post-launch live-ops.
fa-product-money-gate	product	Enforcing the money gate at each stage.
fa-unit-economics-model	product	Building/maintaining the shared net-of-fee economics model.
fa-product-portfolio-retrospective	product	Reviewing outcomes across the portfolio.

---

Install

GitHub Copilot

copilot plugin marketplace add frostaura/ai.toolkit.gaia

copilot plugin install foundation@frostaura
copilot plugin install engineering@frostaura
copilot plugin install product@frostaura

Claude Code

/plugin marketplace add frostaura/ai.toolkit.gaia

/plugin install foundation@frostaura
/plugin install engineering@frostaura
/plugin install product@frostaura

Install only what you need — foundation is the base, engineering adds software delivery, product adds product discovery.

---

Use it

Open any project and prompt your assistant:

"Create a REST API for a blog with posts and comments."

Gaia refines the request, drafts architecture into docs/, plans the work, implements it, tests it, and validates release gates — automatically.

Headless (Copilot CLI)

copilot -p "Create a REST API for a blog with posts and comments" --yolo

Product discovery (one-off / in-app purchases)

The product plugin runs a money-gated pipeline for consumer products that monetize via one-off or in-app / in-game purchases (not SaaS):

/fa-product-coordinator premium one-off puzzle game for iOS + Steam, ~$200k net in year one

…or describe the goal in natural language. The coordinator frames the brief, then runs 10 money-gated stages — each as fan-out → adversarial review → synthesis against one shared unit-economics model — and returns an Opportunity Thesis with a go / loop-back / kill decision.

---

Disclaimers

Gaia uses a remote MCP server by default so plans, memories, and evolution lessons persist across machines and sessions (including GitHub Copilot's web coding agent).

• Stored: evolution suggestions, task plans, project memory — all segregated by project name.
• Not stored: user PII, project code, specs, or documentation.

Prefer fully local? Point the MCP config at a local STDIO server instead. The server source lives in src/Gaia.Mcp.Server.

---

Local development

Working on Gaia itself? Install from a local clone so changes are picked up immediately. Point the install at a plugin directory (where plugin.json lives), not the repo root.

GitHub Copilot

copilot plugin install /absolute/path/to/ai.toolkit.gaia/plugins/foundation

Claude Code

Claude Code clones the marketplace source, so the published .claude-plugin/marketplace.json points each plugin at the remote repo. For local dev, temporarily point the source at the in-repo plugin directory so local edits are picked up:

// .claude-plugin/marketplace.json
"source": "./plugins/foundation"

⚠️ This change is for local development only — do not commit it. Revert to the published remote source before pushing.

Then install from your local clone:

/plugin marketplace add /absolute/path/to/ai.toolkit.gaia
/plugin install foundation@frostaura

MCP server

dotnet build src/Gaia.Mcp.Server/Gaia.Mcp.Server.csproj

---

Contributing

See AGENTS.md. Keep this README concise and current with any new features or changes. Contributions are welcome under the MIT License.

---

"In Greek mythology, Gaia is the personification of Earth and the ancestral mother of all life."