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.

​

Platform features

Use this page as a map of the platform. Start with the spine: one persistence and discovery model backs catalog kinds, the vault, hybrid search, and execution configuration. Neurons, pathways, receptors, and reflexes are resource-backed product objects; receptors are how HTTP and channels invoke them. Endpoint details live in the API reference and guides.

​

Unified resource spine

Everything durable—neurons, pathways, receptors, reflexes, assets, artifacts, stakeholder maps, meetings, skills, and more—is stored as resource heads with versions, optional content-addressed bodies, and chunks used for /v1/search and RAG. You interact with that model in three layers:

1. Discovery — cross-kind lists and trees: Browse, Resources (POST /v1/browse, resource tree, GET /v1/resources/{catalog} where supported).
2. Text and meaning — keyword + vector retrieval over indexed material: Search, Search guide (POST /v1/search, async GET /v1/search).
3. Document lane — immutable assets and versioned artifacts: Assets and Artifacts; SDK resourceStore.

Recollections scope retrieval for neurons and traces over that same indexed corpus—see Recollections. Traces and engrams (conversation history) are a separate plasticity plane from vault bodies; use trace APIs for transcript data, the spine for durable knowledge you want indexed like other resources.

​

Neurons

Neurons are configurable agents: model, prompts, tools, recollections, and scope (core | workspace | brain). Catalog reads and soft-delete use /v1/resources/neurons/...; create, update, invoke, and slug-shaped reads use /v1/neurons—see Neurons API.

• Neuron scope

​

Pathways

Pathways are workflows: validate, compile, and execute through the API.

• Pathways API

​

Execution and invoke

Receptors target a neuron or pathway and handle auth, validation, and invoke/stream semantics. Jobs represent long-running work. POST /v1/neurons/:id/invoke uses the same job type; optional neuron_task_options.trace_type controls background engram listing (Neurons → Invoke neuron).

• Integrate the API — end-to-end from API key to production
• Neurons API — CRUD and invoke (including neuron_task_options)
• Receptors API — list, invoke, share flows
• Jobs API — start runs, status, stream, cancel
• SDK streaming — typed streaming helpers in TypeScript

​

Reflexes and tools

Reflexes group tools (native built-ins and connected integrations). Tools are the callable actions agents and pathways use.

• Reflexes API
• Tools API
• MCP tools — assistant-facing tool surface on the hosted MCP server

​

Knowledge

Assets, artifacts, and recollections attach indexed knowledge to agents and traces via the spine.

• Assets and Artifacts API
• Recollections API

​

Search and discovery

Search (/v1/search) combines keyword matching on indexed chunks with optional vector retrieval (default hybrid). List-by-type discovery uses Browse.

• Search API
• Search guide — patterns and examples

​

Observability

Traces carry multi-turn conversation and execution context. Activity exposes engram-style audit trails; brain scoping applies for some routes—see the reference.

• Traces API
• Activity API

​

How to connect

Pick the surface that fits your stack; the underlying spine is the same.

​

Related

• Authentication — API key, JWT, share token, MCP
• API reference overview — base URL, errors, spine vs vertical routes
• Integration patterns — architectural patterns for apps