Naftiko turns your existing data and APIs into a declarative capabilities exposing MCP, Agent Skills and REST servers.
You don't replace your APIs. You don't rip-and-rebuild for AI. You take stock of what already exists, organize it into capabilities, and make it usable by humans and machines across every domain in your business.
Our documentation hub is the developer front door: it explains the Spec-Driven Integration (SDI) approach, gets you started with zero installation, routes you into the right tutorial track depending on your use case.
- APIs are everywhere, but not always being used
- Specs exist, but no shared meaning
- MCP servers popping up in ad hoc ways
- Copilots and agents lack the context they need
- Agent Skills seem like the solution, without fully knowing why
- Leadership mandates AI, teams are absorbing the risk
SDI is the methodology at the core of Naftiko. Every integration is a declarative specification first — authored once in YAML, validated by a deterministic linter, executed by a deterministic engine, governed by deterministic policy, and orchestrated across the Fleet. One spec, six tools, every protocol surface (MCP + Skill + REST) from the same definition.
Developers only need to know YAML, JSONPath, and Mustache templates to define capabilities — no Java required unless extending the engine itself.
| Feature | Description |
|---|---|
| Spec-Driven | Declare capabilities entirely in YAML |
| Multi-Protocol Servers | Expose capabilities via MCP, Skill, and REST from the same spec |
| Data Format Conversion | Transform Protobuf, XML, YAML, CSV, and Avro payloads into JSON |
| HTTP API Consumption | Connect to any HTTP-based API with built-in authentication support |
| Templating & Querying | Mustache templates and JSONPath expressions for flexible data mapping |
| AI Native | Built for context engineering and agent orchestration from the ground up |
| Docker Native | Ships as ready-to-run Docker containers |
| Extensible | Open-source core extensible with new protocols and adapters |
Our documentation hub organizes its tutorials around the four pains the Fleet exists to solve:
- Track 1 — Context Engineering — Design for MCP, then wire the APIs.
- Track 2 — API Reusability — Turn old API investment into new experiences.
- Track 3 — API Orchestration — Aggregating multiple source APIs in a domain-driven way.
Each track answers one of the four pains: API sprawl, AI agents drifting / hallucinating / breaking contracts, context fragmentation for AI tasks, and integration boilerplate.
- Maximize Existing Investments — Your data and APIs are not technical debt, they are your strategic inventory.
- Capabilities, Not Just Endpoints — AI doesn't need endpoints, it needs capabilities that are spec-driven.
- Governed AI Integration at Scale — AI integration without governance does not scale and does not survive.
- Meet Teams Where They Are — Reusability shows up in the IDE; governance is seamless guidance.
- Open by Default — Ikanos and Polychro are Apache 2.0 in every edition. Signals is open data, versioned in Git.
Naftiko operates from Paris and Philadelphia.