Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.auvy.ai/llms.txt

Use this file to discover all available pages before exploring further.

AUVY Docs

Most durable product data lives on a single unified resource spine: versioned heads in Mongo (GET /v1/resources/... and related catalog routes), document lane rows as assets (ingested) and artifacts (mutable), and lexical + vector discovery over resource_chunks via /v1/browse, /v1/search, and /v1/assets / /v1/artifacts. Execution—invoking agents and workflows—still flows through neurons, pathways, and receptor gateways; those objects are resource-backed like everything else. Integrate with an API key (server), JWT + publishable key (browser), or the hosted MCP server (assistants). The same platform is available as:
  • REST API
  • @auvy-synapse/client
  • CLI
  • MCP server

Unified resource spine — how to use it

GoalTypical API / SDK surface
List or page a kind (neurons, pathways, receptors, …)POST /v1/browse, or catalog lists such as GET /v1/resources/neurons — see Resources and Browse
Hybrid or keyword discovery (type/pattern queries)POST /v1/search (sync) or GET /v1/search (async job) — Search, Search guide
Vault / folder navigationPOST /v1/browse/resource-tree, GET/POST /v1/resources/treeResources
Ingest or update document lanePOST /v1/assets/ingest, artifact PATCH/patch envelopes — Assets and Artifacts
Symmetric catalog read / soft-delete / restore (e.g. neuron UUID)GET / DELETE / restore under /v1/resources/{catalog}/:id — API reference per kind
Product-shaped CRUD + invoke (slug neurons, receptor invoke, pathway execute)/v1/neurons, /v1/receptors, /v1/pathways — same rows as the spine, different route shapes
In the TypeScript SDK, auvy.resources is the main cross-kind entry point; auvy.resourceStore covers assets and artifacts. See SDK resources.

What you can build

Platform features

Spine-first map: discovery, document lane, synapse objects, receptors, and observability.

Integrate the API

Quick start and end-to-end: API key to production streaming.

Neuron scope

Core, workspace, and brain scope—and how it relates to receptors.

Authentication

API key, browser, CLI, MCP, and share-token flows.

TypeScript SDK

Public client setup, auth modes, resources, and streaming helpers.

API Reference

Endpoints, payloads, responses, and patterns.

Choose your entry point

REST API

Call https://api.auvy.ai/v1 directly from any language.

TypeScript Client

Use @auvy-synapse/client for typed requests, retries, and streaming helpers.

CLI

Script common workflows from the terminal with auvy.

MCP Server

Connect AUVY to Cursor, Claude Desktop, ChatGPT, and other MCP clients.

What you need to know

  • Resource spine — one model for heads, versions, document lane, and search indexes; use Resources / Browse / Search before ad-hoc per-kind guessing.
  • Neurons are the primary agent object you configure; scope is core, workspace, or brain (see Neuron scope).
  • Receptors are gateways: stable URLs and channel bindings that invoke a neuron or pathway.
  • Most invoke examples use POST /v1/receptors/:workspace_slug/:slug_or_id/invoke.
  • Long-running work returns a job_id to poll or stream.
  • Server-side usage should use an API key.
  • Browser usage should use a publishable key plus a user JWT.
  • Shared/public flows use a receptor share token.
  • JWT-only workspace integration management is intentionally out of scope for this docs set.
1

Understand the spine (optional)

Skim Platform featuresUnified resource spine first, then execution and tools.
2

Create an API key

Generate a key in the dashboard, then keep it server-side.
3

Discover and store context

Use Browse or Search for catalog discovery; use Assets and Artifacts when you need durable documents for RAG.
4

Define a neuron, then invoke

Create or select a neuron (dashboard or POST /v1/neurons), ensure a receptor targets it, then call POST /v1/receptors/.../invoke or use the SDK.
5

Add polling or streaming

Use the Jobs API or the client streaming helpers for long-running responses.
6

Add CLI, MCP, or Connect when needed

Use the CLI for scripts, the hosted MCP server for assistant workflows, and Connect when you need external integrations or account connections.

Contributors

Working in the Synapse monorepo? Use the Developer tab: Developer overview, Local development, and Documentation site (how to edit this Mintlify app, validate, and ship doc changes).

Next steps

  1. Platform features for spine + feature map with deep links.
  2. Resources API and Search guide for day-to-day spine usage.
  3. Integrate the API for quick start and full walkthrough.
  4. Neuron scope for tenancy and core-admin rules.
  5. Authentication to choose the right credential model.
  6. TypeScript SDK, CLI Reference, or MCP Server for your preferred surface.