Memory API

The Memory API exposes developer-facing retrieval, encoding, event ingestion, control, provenance, and tombstone primitives under /v1/memory. App-specific workspace and admin surfaces remain under /api.

Use this section as the practical map for authentication, SDK setup, memory-augmented inference, signed-in workspace memory views, candidate review, knowledge memory, reflection, preference calibration, and habit controls.

Authentication

Surface	Auth	Scope
/v1/* inference gateway	Bearer MCP access token such as acm_...	Token must include inference:chat
/v1/memory/retrieve	Bearer MCP access token such as acm_...	Token must include memory:read
/v1/memory and /v1/memory/events	Bearer MCP access token such as acm_...	Token must include memory:write
/v1/memory/:id/reinforce, /v1/memory/:id/suppress, and /v1/memory/:id/provenance	Bearer MCP access token such as acm_...	Token must include memory:control
DELETE /v1/memory/:id	Bearer MCP access token such as acm_...	Token must include memory:delete
/api/user/memory/*	Signed-in workspace session	Current user in the current organization
/api/assistants/:id/memory/*	Signed-in workspace session	Assistant access and memory permissions
/api/organizations/:id/memory/*	Signed-in workspace session	Organization owner or admin
/api/memory/*	Signed-in workspace session	Memory permissions required by route

API surfaces

• Developer Memory API: public /v1/memory retrieve, encode, event-ingestion, control, provenance, and tombstone endpoints.
• SDK: install and use @achiral/chiro from Node, Next.js, Express, Workers, and Vercel AI SDK projects.
• Event ingestion: send deployments, incidents, CI runs, PRs, support tickets, and preference events into Chiro memory.
• Inference gateway: OpenAI-compatible /v1 chat completions for memory-augmented inference.
• User memory: signed-in account memory summary and personal memory candidate review.
• Assistant memory: assistant memory summary and assistant-scoped candidate review.
• Organization memory: organization memory distribution, member summaries, and organization candidate review.
• Core memory candidates: approve or reject flashbulb candidates before they become durable core memory.
• Knowledge memory: create, search, update, and tombstone organization knowledge records.
• Reflection loop: configure, inspect, and run reflection over memory.
• Preference calibration: feedback and calibration routes for learned preferences.
• Habit memory API: personal controls for Chiro's habit behavior.

Continue from here

• API reference for shared API conventions, base URLs, streaming, and errors.
• ACT-R memory for the product model behind activation, retrieval, decay, and consolidation.