tama_mem

Memory tools provide a shared key-value store accessible to all agents within a single run. Use them to pass data between agents in a pipeline without writing to disk.

Memory is in-process and ephemeral — it does not persist across runs.

tama_mem_set

Stores a string value under a key.

Parameters

Name	Type	Description
key	string	Storage key, e.g. "summary"
value	string	Value to store

Example skill

---
name: mem-set
description: Store a value in shared pipeline memory so other agents can retrieve it.
tools: [tama_mem_set]
---

Use tama_mem_set(key, value) to store a string under a key.
Other agents in this pipeline can retrieve it with mem-get.

tama_mem_get

Retrieves a value stored by a previous agent.

Parameters

Name	Type	Description
key	string	Storage key to retrieve

Returns the stored value, or [no value stored for key '...'] if not set.

Example skill

---
name: mem-get
description: Retrieve a value stored by another agent earlier in the pipeline.
tools: [tama_mem_get]
---

Use tama_mem_get(key) to retrieve a value stored with mem-set.

tama_mem_append

Appends an item to a JSON array stored under a key. Creates the array if it doesn’t exist.

Parameters

Name	Type	Description
key	string	Array key, e.g. "results"
item	string	JSON-encoded value to append

Returns the full array after appending.

Example skill

---
name: mem-append
description: Append an item to a shared array in pipeline memory.
tools: [tama_mem_append]
---

Use tama_mem_append(key, item) where item is a JSON-encoded value.
Examples:
- tama_mem_append("errors", '"something failed"')
- tama_mem_append("results", '{"agent": "analyst", "score": 0.9}')

Read the full array with tama_mem_get(key).

Pipeline pattern

A typical multi-agent pipeline uses mem to pass state:

agent-a  → tama_mem_set("report", <text>)
agent-b  → tama_mem_get("report") → processes it
agent-c  → tama_mem_append("findings", <json>)
reporter → tama_mem_get("findings") → synthesizes