Agentic Memory Module — MCP Tool Specification

This document describes the 3 MCP tools exposed by codomyrmex.agentic_memory.mcp_tools.

Note: The memory_put, memory_get, and memory_search tools use a fresh InMemoryStore per invocation. They are designed for single-session in-process use. For persistent memory across invocations, use the Python API with JSONFileStore directly.

---

memory_put

Store a new memory entry with content, optional type, and importance.

Input Schema

Parameter	Type	Required	Description
content	str	Yes	Text content of the memory
memory_type	str	No	One of "episodic", "semantic", "procedural" (default "episodic")
importance	str	No	One of "LOW", "MEDIUM", "HIGH", "CRITICAL" (default "medium")

Output Schema

{
  "id": "550e8400-e29b-41d4-a716-446655440000",
  "content": "The project deadline is March 1st",
  "memory_type": "episodic",
  "importance": 2,
  "metadata": {},
  "tags": [],
  "created_at": 1740340308.123,
  "access_count": 0,
  "last_accessed": 0.0
}

---

memory_get

Retrieve a memory by its ID.

Input Schema

Parameter	Type	Required	Description
memory_id	str	Yes	UUID of the memory to retrieve

Output Schema

Returns the memory dict (same format as memory_put output), or null if not found.

{
  "id": "550e8400-e29b-41d4-a716-446655440000",
  "content": "The project deadline is March 1st",
  "memory_type": "episodic",
  "importance": 2,
  "metadata": {},
  "tags": [],
  "created_at": 1740340308.123,
  "access_count": 1,
  "last_accessed": 1740340320.456
}

---

memory_search

Search stored memories by a text query and return ranked results.

Input Schema

Parameter	Type	Required	Description
query	str	Yes	Text query for relevance scoring
k	int	No	Maximum results to return (default 5)

Output Schema

[
  {
    "memory": {
      "id": "550e8400-e29b-41d4-a716-446655440000",
      "content": "The project deadline is March 1st",
      "memory_type": "episodic",
      "importance": 2,
      "metadata": {},
      "tags": [],
      "created_at": 1740340308.123,
      "access_count": 0,
      "last_accessed": 0.0
    },
    "relevance": 0.75,
    "combined_score": 0.52
  }
]

Results are sorted by combined_score descending (0.5*relevance + 0.3*recency + 0.2*importance).