Need Orchestration Tooling for Fleets of Role-Based Agents
Need tooling that lets a non-engineering operator define skills, create role-based agents, give each agent its skill, and wire them together as a working fleet — VP of Marketing agent, Social Media Manager agent, Paid Ads Manager agent — on a single reviewable dashboard.
Take Control Of Your Signals — Become a Naftiko Design Partner Today!
Persona Story:
Harper is already running her marketing function as a fleet of role-based agents — a VP-of-Marketing agent, a Social-Media-Manager agent, a Paid-Ads-Manager agent — each with skills, all reporting cron-generated tasks into a vibe-coded review dashboard. The framing is structural: if she were building a physical team, what would the job specs be? Then transpose that into skills for agents. The missing layer is the orchestration tooling itself — define a skill, create an agent, give it that skill, then wire it all together and define workflows between roles.
Problem Context
- Non-engineering function leads (product marketing, ops, sales) are already standing up fleets of role-based agents in production
- The job-spec-as-agent-spec frame is real: practitioners think in roles first, then translate roles into skills for agents
- Current tooling forces operators to glue together Claude agents, cron jobs, and bespoke dashboards by hand
- The shape practitioners describe is exactly: “define a bunch of skills, create an agent, give it that skill, then a way of just wiring it together and defining workflows. But I don’t know whether they’re anything.”
Problem Impact
- Smaller teams scale by identifying which roles can be AI and which need humans, but burn weeks bolting the result together
- Vibe-coded dashboards are not auditable, governable, or shareable across functions
- Orchestration across role-based agents stays ad hoc — there is no consistent workflow primitive
- Every team that adopts the pattern reinvents it, so the lessons do not compound across the company
Naftiko Today
- Executable YAML capability specs are the natural skill-definition surface — each capability is a skill an agent role can be assigned
- Agent Skills exposure makes capabilities directly assignable to specific agent roles
- MCP exposure (Streamable HTTP, stdio) lets each role-based agent consume its assigned skills through a single canonical surface
- External bindings keep credentials and scopes per-role, so the VP-of-Marketing agent and the Paid-Ads-Manager agent operate inside their own authorization perimeters
Naftiko Tomorrow
- Naftiko Shipyard MVP (Fleet Second Alpha) will be the workbench Harper assembles a fleet on — roles, skills, wiring, review dashboard, all in one surface
- Starter templates (Second Alpha) will publish reference role-based agent fleets (marketing, operations, support) so operators start from a working pattern
- VS Code Extension (Fleet) and Backstage Integration (Fleet) will bring role-based-agent orchestration into the surfaces engineering teams already use
- A2A (Agent-to-Agent) adapter (Second Alpha) will define how role-based agents communicate inside a fleet without each operator inventing the wiring