Need API-First Internal Software with MD-File Agent Instructions
Internal enterprise software is shifting to a pattern where every service is an API, every API is exposed to agents via markdown / skills files, and users themselves activate or deactivate features — replacing generic SaaS with customer-specific agentic interfaces.
Take Control Of Your Signals — Become a Naftiko Design Partner Today!
Persona Story:
Maya is shipping internal enterprise software in a recognizably new shape: every feature is an API, agents are trained through markdown or skills files to use those APIs, and everyone participates in the development of the solution and the features over time. Users themselves bring in features they want and activate or deactivate features they do not. The generic-SaaS layer is dissolving — what replaces it is customer-specific software organized around APIs and agent-readable instructions.
Problem Context
- Forward-deployment AI engineers inside enterprises name this pattern in customer words: every service an API, every API agent-readable via markdown, users activating their own features
- Generic SaaS is failing because it had to build platforms everyone could use but was never specific enough — custom internal software with agent interfaces is the replacement
- The pattern is architectural — feature surfaces, agent instructions, user controls — not a single technology choice
- The pattern has to be reproducible across customers, not bespoke per engagement
Problem Impact
- Without a canonical pattern, every forward-deployment engagement reinvents the wheel
- Customers cannot self-modify their own internal software because the feature surface is hidden behind code
- Agents need bespoke training per customer because each implementation organizes its APIs and instructions differently
- The promise — users participating in feature development — stays unrealized at scale
Naftiko Today
- Executable YAML capability specs realize “every service is an API” as a deterministic, declarative artifact rather than a hand-written API contract
- Agent Skills exposure publishes capability metadata as agent-readable skills, so the “agents trained through markdown / skills files” pattern is the default
- MCP exposure layer (Streamable HTTP, stdio) lets every internal-software feature be addressable by any MCP-compatible agent
- Capability YAML’s parameter and binding model lets users toggle which features are active without code changes
Naftiko Tomorrow
- HTML / Markdown format support (Second Alpha) will render capabilities as the markdown agent-instruction files this pattern depends on
- Tool annotations (Second Alpha) will let internal-software builders signal which features are user-toggleable and which are administrative
- Starter templates (Second Alpha) will publish reference implementations of the API-first-internal-software pattern so forward-deployment engineers stop reinventing it
- Fabric capability discovery (v1.1) will make every internal-software feature discoverable across the enterprise capability fabric