Ainfera builds the routing layer for AI agent inference. Every request is routed to the optimal model based on quality, cost, and latency constraints — and every response carries a cryptographically signed audit receipt that can be independently verified offline.
Trust no one — verify yourself.
- Routing runtime — L2 inference routing with policy templates, audit chain, and health-checks
- Verification — Offline audit chain verifier. No network calls. No trusted third parties
- SDK — Python SDK with signed AgentCards, routed inference, and audit chain
- MCP integration — Model Context Protocol adapter for Claude, Cursor, and agent frameworks
- Open methodology — Reproducible routing benchmarks and stable API contracts (CC-BY 4.0)
- Research — Open science for outcome-aware inference routing
| Repository |
Description |
| api |
L2 routing runtime — /v1/inference, workflows, audit chain |
| sdk-python |
Python SDK — signed AgentCards, routed inference, audit chain |
| routing |
Policy templates + routing methodology for production agents |
| verify |
Offline audit chain verifier — trust no one, verify yourself |
| Repository |
Description |
| mcp |
MCP reference adapter for Claude Agent SDK / Cursor |
| examples |
Integration examples — LangChain, CrewAI, LangGraph, OpenAI, Hermes |
| specs |
Open API contracts + benchmark harness (CC-BY 4.0) |
| Repository |
Description |
| research |
Open science for outcome-aware inference routing |
| Repository |
Description |
| web |
Marketing site + dashboard (ainfera.ai) |
| Repository |
Description |
| labs |
Offline-learning deploy — catalog intake, replay, methodology |
| valinor |
Agent-fleet substrate + MCP gateway |
| vault |
Company knowledge vault — strategy, architecture, governance |
- Every repo has a real README with setup + usage instructions
- Public claims are derived from live API state — no hardcoded model counts
- Public-facing claims in
web are mapped to API endpoints or signed fixtures
- All public repos licensed under Apache 2.0 unless stated otherwise
- Security reports via SECURITY.md or security@ainfera.ai
All public repositories are licensed under Apache 2.0 unless otherwise stated in the repository.
For security reports, see SECURITY.md or email security@ainfera.ai.
