mesa_llm/memory/memory.py (2)

56-57: Minor formatting inconsistency for non-dict list items.

Line 57 appends a trailing colon after non-dict items ({item} :), but since these are values rather than key-value pairs, the colon appears unnecessary and inconsistent with how scalar values are displayed elsewhere (line 61 also has this pattern for non-list scalars).

💡 Suggested fix

                     else:
-                        lines.append(f"      [blue]└──[/blue] [cyan]{item} :[/cyan]")
+                        lines.append(f"      [blue]└──[/blue] [cyan]{item}[/cyan]")

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In `@mesa_llm/memory/memory.py` around lines 56 - 57, The list branch currently
appends non-dict items with a trailing colon using the lines.append call that
builds f"      [blue]└──[/blue] [cyan]{item} :[/cyan]"; remove the colon so
scalar list items render like other scalars (match the pattern used elsewhere
for non-list scalars, e.g., the similar append at the non-list scalar branch),
updating the string to omit " :" and ensure whitespace/formatting matches other
scalar lines.

---

143-146: Type annotation mismatch: frozenset assigned to set[str].

The annotation declares set[str] but the value is a frozenset. While this works at runtime, it's technically incorrect for static type checkers and could mislead subclasses about mutability.

💡 Suggested fix

-    ADDITIVE_EVENT_TYPES: set[str] = frozenset({"message", "action"})
+    ADDITIVE_EVENT_TYPES: frozenset[str] = frozenset({"message", "action"})

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In `@mesa_llm/memory/memory.py` around lines 143 - 146, The annotation for
ADDITIVE_EVENT_TYPES is incorrect: it declares set[str] but is assigned a
frozenset; update the type to match immutability by changing the annotation to
frozenset[str] (or typing.FrozenSet[str]) so ADDITIVE_EVENT_TYPES:
frozenset[str] = frozenset({"message", "action"}) and keep the existing value;
this ensures static type checkers and readers see the correct immutable type for
the constant.

mesa_llm/memory/st_memory.py (1)

97-111: LGTM with a consistency note.

The implementation correctly handles both list-valued and scalar message entries. However, EpisodicMemory.get_communication_history() (lines 150-160) directly accesses entry.content['message'] without checking if it's a list, unlike the updated approach here. If both classes are used with additive message storage, they should handle messages consistently to avoid formatting errors.

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In `@mesa_llm/memory/st_memory.py` around lines 97 - 111,
EpisodicMemory.get_communication_history currently assumes
entry.content['message'] is a scalar; change it to match the logic used in
get_communication_history of the short-term memory: first check that "message"
exists in entry.content, retrieve msgs = entry.content["message"], then if
isinstance(msgs, list) iterate and append each msg with the same formatting
(e.g., f"step {entry.step}: {msg}\n\n") else append the scalar message with the
same formatting; ensure you reference EpisodicMemory.get_communication_history
and use entry.step and entry.content consistently.