cencori weekly downloads statistics

Cencori

The unified infrastructure layer for AI applications.

One SDK. Every AI primitive. Always secure. Always logged.

Quick Start

Products

| Product | Status | Description |

| AI Gateway | ✅ Available | Multi-provider routing, security, observability | | Web | ✅ Available | First-party fetch, extraction, crawl, and evidence-bearing search | | Billing | ✅ Available | End-user usage monetization and Stripe integration | | Integration | ✅ Available | SDKs, Vercel AI, TanStack | | Embedded Agents | ✅ Available | Multi-tenant agent backend: tenants, versions, runs, approvals, metering |

Embedded Agents

Namespaces: tenants, clientTokens, models, providerConnections, agentVersions, installations, runs, actions, knowledge, connections, mcpServers, skills, skillImports, webhooks, usage, endUsers, ratePlans. Full reference: https://cencori.com/openapi/embedded-agents.json.

AI Gateway

Chat Completions

Multimodal (Image Input)

Tool Usage (Function Calling)

Structured Output (Object Generation)

Embeddings

Vision (Image Understanding)

Analyze, describe, OCR, and classify images. Routes across OpenAI, Anthropic, and Google vision-capable models.

Drop-in React uploader for your own product (cencori/react):

Full API in docs.

Judgment (Structured Decisions)

Yes/no, choice, and score answers with probabilities over any state — no generated text. Any judgment catalog model works here (TypeSafe's Jev today).

Full API in docs.

Documents (PDF & Image Extraction)

Extract text from PDFs and images, summarize documents, and answer questions about them. Text-based PDFs use native parsing — no LLM tokens, free.

Full API in docs. See the Contract Analyzer tutorial for an end-to-end example.

Web Search and Retrieval

Fetch and extract public pages, build a private project corpus, or search Cencori's continuously refreshed public index:

For tool-calling agents, import WEB_SEARCH_TOOL or WEB_FETCH_TOOL from cencori/web, then resolve emitted calls with cencori.web.executeTool(...). Retrieved page content is always marked as untrusted input.

Image Generation

Generate images from text prompts using multiple providers:

Supported Models: | Provider | Models | Description |

| OpenAI | gpt-image-1.5, gpt-image-1, dall-e-3, dall-e-2 | Text rendering, creative | | Google | gemini-3-pro-image, imagen-3 | High photorealism |

Framework Integrations

Vercel AI SDK

With React/Next.js

Why Cencori?

• 🛡️ Security Built-in: PII detection, content filtering, jailbreak protection • 📊 Observability: Every request logged, every token tracked • 💰 Cost Control: Budget alerts, spend caps, per-request costing • 🔄 Multi-Provider: Switch between OpenAI, Anthropic, Google, etc. • ⚡ One SDK: AI model routing, security shield, and billing - unified

Links

• Documentation • Dashboard • GitHub

License

MIT