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.

​
MCP Tools Reference

The hosted MCP server (apps/mcp only) merges tools from the active workspace MCP toolkit (GET /v1/mcp-toolkits/resolve; optional headers X-MCP-Toolkit-Id, X-MCP-Toolkit-Slug): native tools (same names and schemas as GET /v1/tools where included), optional management/search/CRUD tools, optional connect_execute pairs, and optional receptor__{slug} tools. Configure toolkits in the dashboard or via /v1/mcp-toolkits.

​
Session Context

​
get_session

Returns connection context for the authenticated user:
  • workspace_id
  • workspace_slug
  • brain_id
  • default_brain_id
Use this first to verify workspace/brain scope.

​
Discovery Tools

​
grep

Keyword browse/search using query in type/pattern format. List a type with type or type/ (assets, artifacts, receptors, pathways, neurons, reflexes); filter with neurons/support-style patterns. For assets/ and artifacts/, results use the resource row UUID in path (assets/<uuid>, artifacts/<uuid>) and may include kind, content_type, native_path, body_policy—same shape as search. User queries may include memories/; the server normalizes that token to assets/ for path display. List-only by type (no pattern) is also available via the SDK browse resource (GET/POST /v1/browse). Semantic search using query in type/pattern format. For assets/… vs artifacts/…, the API scopes to that kind (ingested assets vs authored artifacts). Results can include kind, content_type, native_path, and body_policy alongside path. Use read for the document JSON; use raw body export on the assets lane (GET /v1/assets/:id/body where configured). Other types (neurons, pathways, …) follow the same path shape as grep.

​
read / cat

read takes resource_id only (one unified spine UUID—same as document_tree rows[].id or search). cat is an alias and uses the same shape. For engrams and transcript history, use read_history with engrams/<id> paths instead of read.

​
grep_history / read_history

Inspect prior execution history for the current neuron and brain context. grep_history accepts optional query (keyword filter) and optional limit (integer 1–100, default 100). Results include path values like engrams/<id> for use with read_history.

​
list_neuron_reflexes

List reflexes for a neuron. Required: neuron_path (e.g. neurons/support-agent or neurons/<uuid>).

​
Web Tools

​
web_read

Fetch page content from one or more URLs. Run one or more web searches.

​
Creation And Update

​
write_reflex

Creates a reflex via native create_resource under the hood: pass reflex fields at the top level with required name (same schema spirit as POST /v1/reflexes). Toolkit must include create_resource.

​
apply_patch

Apply a ResourcePatchEnvelope to an existing catalog resource: kind, resource_id, patch. HTTP: POST /v1/resources/{catalog}/{resource_id}/patch. Use artifact_apply_text for UTF-8 artifact bodies, or domain-specific *_partial modes.

​
apply_artifact_patch

Narrow artifact UTF-8 helper — POST /v1/artifacts/patch with the patch envelope string (*** Begin Patch … *** End Patch). Target lines after *** Write File: / *** Update Artifact: use the unified artifact resource id (UUID); prefer UUIDs over path-shaped artifacts/<uuid> lines in authored patches.

​
create_asset

Immutable asset text ingest — POST /v1/assets/ingest via native create_asset.

​
create_resource

Create synapse/pathway/reflex/receptor resources by path + config (e.g. pathways/foo, neurons/foo, reflexes/foo, receptors/foo). Neurons require config.neuron_config; receptors require config.target_id (+ optional config.target_type).

​
Document pipeline (summary)

Mutable catalog bodies and metadata use apply_patch / vertical PATCH routes on the unified spine—prefer these over ad-hoc native edit / store tools where a toolkit exposes them.
SurfaceNotes
Artifact UTF-8apply_patch (artifact_apply_text) or apply_artifact_patch
AssetsNew rows: create_asset (immutable ingest)
Neurons / pathways / reflexes / receptorscreate_resource or management CRUD; updates apply_patch where enabled
Raw export (artifacts): GET /v1/artifacts/:id/body (lane-scoped UTF-8 export).

​
Execution

​
delegate

Delegate to a neuron by path (neurons/<name>). This calls POST /v1/receptors/invoke with target_path. Receptor-specific tools may appear as receptor__{slug} from the toolkit resolver—those hit /v1/receptors/invoke with a receptor slug.

​
connect_execute

Execute a Connect integration tool directly. AUVY uses direct execution: you choose which tools to run and call connect_execute with the integration and tool name (no session or meta-tool discovery). Required input:
  • integration_id (UUID)
  • tool_name
Optional input:
  • tool_config (object) — tool arguments
  • connected_account_id — which connected account to use when the user has multiple
Use connect_execute with an integration that has already been configured in the AUVY dashboard. This docs set does not cover the separate JWT-only integration-management surface.

​
Skills

​
learn_skill

Load a skill by skill_id (UUID or skill path name). Returns name, description, and body for use at runtime.

​
remove_neuron_reflex

Remove a neuron-reflex association. Required: neuron_path and reflex_path (e.g. neurons/support-agent, reflexes/<uuid> or reflexes/<name>—same segments as read).

​
MCP App UI

The read tool can return an MCP App resource for receptors:
  • ui://receptor-viewer/mcp-app.html
The current MCP app is implemented in Vue (apps/mcp/resources/receptor-viewer).

​
Brain Context

Some tools are brain-scoped. Set X-Brain-Id in MCP headers when needed: 
{
  "headers": {
    "Authorization": "Bearer ak_live_xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx",
    "X-Brain-Id": "00000000-0000-0000-0000-000000000000"
  }
}