Types

MemoryUnit

The atomic unit of knowledge in the Akashik Protocol.

MemoryUnit

The MemoryUnit is the atomic unit of knowledge in the protocol. Every finding, decision, observation, question, or intention recorded by an agent is a MemoryUnit. It is immutable once committed — a committed unit's content MUST NOT be modified. To update or correct information, an agent records a new unit with a supersedes or correction relation pointing to the original.

Two properties distinguish Akashik memory units from generic data storage:

Intent is required. Every unit MUST include an intent field declaring why it was recorded. A unit without intent is invalid and MUST be rejected by the Field.
Confidence is optional, and never authoritative. An agent MAY attach a confidence score and a short reason to a unit. When it does, the Field MUST carry that confidence unchanged and surface it on every read path — but MUST NOT use it to rank relevance, weight recency, or auto-resolve a conflict. Confidence is an input to the reading agent's judgment, not the protocol's.

Schema

{
  "id": "<string>",
  "mode": "draft | committed",
  "type": "<MemoryType>",
  "content": "<string>",
  "intent": {
    "purpose": "<string>",
    "task_id": "<string | null>",
    "question": "<string | null>"
  },
  "confidence": {
    "score": "<number>",
    "reason": "<string>",
    "evidence": ["<string>"],
    "assumptions": ["<string>"]
  },
  "source": {
    "agent_id": "<string>",
    "agent_role": "<string>",
    "session_id": "<string | null>",
    "timestamp": "<ISO 8601>"
  },
  "relations": ["<Relation>"],
  "status": "active | draft | superseded | retracted | contested | pending_enrichment",
  "epoch": "<number>"
}

Field Reference

Field	Type	Required	Description
`id`	string	✓	Auto-generated by the Field. Agents MUST NOT set this.
`mode`	string	✓	`"draft"` or `"committed"` — affects validation rules.
`type`	string	✓	Semantic category. See MemoryType below.
`content`	string	✓	The knowledge payload. Must be non-empty.
`intent.purpose`	string	✓	Why this unit was recorded. REQUIRED at all levels and modes.
`intent.task_id`	string or null		The task this unit serves.
`intent.question`	string or null		The question this unit answers or asks.
`confidence.score`	number		Certainty between `0.0` and `1.0`. OPTIONAL at every level — never required.
`confidence.reason`	string		Why this confidence score was assigned.
`confidence.evidence`	string		Sources supporting the content. Part of the Extended Confidence form — only meaningful once something downstream reads it (Level 2+).
`confidence.assumptions`	string		What is assumed to be true for this unit to hold. Extended Confidence form; see above.
`source`	object	✓	Auto-generated by the Field. Agents MUST NOT set this.
`relations`	Relation		Links to other memory units. See Relation below.
`status`	string	✓	Auto-set by the Field. See status values below.
`epoch`	integer	✓	Logical clock value. Auto-assigned by the Field.

Required Fields by Mode and Level

Field	Draft (Level 0+)	Committed (Level 0+)
`type`	REQUIRED	REQUIRED
`content`	REQUIRED	REQUIRED
`intent.purpose`	REQUIRED	REQUIRED
`confidence.score`	OPTIONAL	OPTIONAL
`confidence.reason`	OPTIONAL	OPTIONAL
`relations`	OPTIONAL	OPTIONAL

Confidence is never required, at any level — see the note on the protocol's confidence principle above.

The Field MUST auto-generate id, epoch, source.agent_id, source.agent_role, source.timestamp, and status on every RECORD. Agents MUST NOT set these fields.

Mode Behavior

Draft mode allows lightweight, low-ceremony recording. The Field or a downstream agent MAY enrich a draft into a committed unit by adding confidence and metadata. The Field MUST set status to "draft".

Committed mode enforces full validation. At every level, intent.purpose is the only required field beyond type and content. Confidence is never required — carried when present, absent otherwise. The Field MUST set status to "active".

Status Values

Status	Description
`active`	The memory unit is valid and visible for attunement.
`draft`	Recorded in draft mode. Awaiting enrichment before full use.
`superseded`	Replaced by a newer memory unit via a `supersedes` relation.
`retracted`	Explicitly withdrawn by an agent or human operator.
`contested`	Involved in an unresolved conflict.
`pending_enrichment`	Committed but embedding generation or enrichment failed; queued for retry.

MemoryType

MemoryType is the semantic category of a memory unit. Implementations MUST support at minimum: finding, decision, observation, and question.

MemoryType = "finding" | "decision" | "observation" | "intention"
           | "assumption" | "constraint" | "question" | "contradiction"
           | "synthesis" | "correction" | "human_directive"

Type	Description
`finding`	A discovered fact or data point
`decision`	A choice made by an agent or human
`observation`	A subjective assessment or interpretation
`intention`	A declared plan or next step
`assumption`	Something taken as true without verification
`constraint`	A limitation or boundary condition
`question`	An open question that needs answering
`contradiction`	An explicit flag that two findings conflict
`synthesis`	A combination of multiple findings into a conclusion
`correction`	An explicit correction of a prior memory unit
`human_directive`	Input from a human operator

Implementations MAY extend this set with additional types. Extended types MUST use a namespace prefix (e.g., acme:custom-type).

Relation

A Relation links a memory unit to another, building a knowledge graph inside the Field.

{
  "type": "<RelationType>",
  "target_id": "<string>",
  "description": "<string | null>"
}

RelationType = "supports" | "contradicts" | "depends_on" | "supersedes" | "retracts"
             | "caused_by" | "elaborates" | "answers" | "blocks" | "informs"

Type	Description
`supports`	This unit provides evidence for the target
`contradicts`	This unit conflicts with the target — triggers conflict detection at Level 1+
`depends_on`	This unit assumes the target is true
`supersedes`	This unit replaces the target; target MUST transition to `superseded`. Open to any agent.
`retracts`	This unit withdraws the target; target MUST transition to `retracted`. Author-only — the Field MUST reject with `RETRACT_NOT_AUTHORIZED` if the requesting agent isn't the target's original author.
`caused_by`	This unit exists because of the target
`elaborates`	This unit adds detail to the target
`answers`	This unit answers a question posed by the target
`blocks`	This unit prevents progress on the target
`informs`	This unit provides useful context for the target

When a RECORD includes a relation with type contradicts, the Field MUST create a Conflict object (Level 1+).

JSON Schema

{
  "$schema": "https://json-schema.org/draft/2020-12/schema",
  "$id": "https://akashikprotocol.com/schema/0.1.0/memory-unit.json",
  "type": "object",
  "required": ["id", "mode", "type", "content", "intent", "source", "status", "epoch"],
  "properties": {
    "id": { "type": "string", "minLength": 1 },
    "mode": { "type": "string", "enum": ["draft", "committed"] },
    "type": {
      "type": "string",
      "enum": ["finding", "decision", "observation", "intention", "assumption", "constraint", "question", "contradiction", "synthesis", "correction", "human_directive"]
    },
    "content": { "type": "string", "minLength": 1 },
    "intent": {
      "type": "object",
      "required": ["purpose"],
      "properties": {
        "purpose": { "type": "string", "minLength": 1 },
        "task_id": { "type": ["string", "null"] },
        "question": { "type": ["string", "null"] }
      }
    },
    "confidence": {
      "type": "object",
      "properties": {
        "score": { "type": "number", "minimum": 0.0, "maximum": 1.0 },
        "reason": { "type": "string", "minLength": 1 },
        "evidence": { "type": "array", "items": { "type": "string" } },
        "assumptions": { "type": "array", "items": { "type": "string" } }
      }
    },
    "source": {
      "type": "object",
      "required": ["agent_id", "agent_role", "timestamp"],
      "properties": {
        "agent_id": { "type": "string" },
        "agent_role": { "type": "string" },
        "session_id": { "type": ["string", "null"] },
        "timestamp": { "type": "string", "format": "date-time" }
      }
    },
    "relations": {
      "type": "array",
      "items": {
        "type": "object",
        "required": ["type", "target_id"],
        "properties": {
          "type": {
            "type": "string",
            "enum": ["supports", "contradicts", "depends_on", "supersedes", "retracts", "caused_by", "elaborates", "answers", "blocks", "informs"]
          },
          "target_id": { "type": "string" },
          "description": { "type": ["string", "null"] }
        }
      }
    },
    "status": {
      "type": "string",
      "enum": ["active", "draft", "superseded", "retracted", "contested", "pending_enrichment"]
    },
    "epoch": { "type": "integer", "minimum": 0 }
  }
}

Specification v0.1.0

The Akashik Protocol - A Shared Memory and Coordination Protocol for Multi-Agent AI Systems

Agent

An autonomous entity that interacts with the Field.