fix: render readable text in get_communication_history for all memory…

[Harsh-617]

Fixes #193

Summary

• get_communication_history() in all three memory classes (ShortTermMemory, STLTMemory, EpisodicMemory) was rendering raw Python dicts instead of readable message strings in the LLM prompt
• Added a shared _format_message_entry() helper in memory.py that extracts human-readable text from the nested dict structure produced by speak_to
• Applied the fix across all three memory implementations and added regression tests

Root Cause

The speak_to tool stores messages as a nested dict:

add_to_memory(
    type="message",
    content={"message": "meet me at the north", "sender": 7, "recipients": [1, 2]},
)

This means entry.content["message"] is always a dict, not a string. But get_communication_history() passed it directly into the f-string:

# Before — renders raw dict
f"step {entry.step}: {entry.content['message']}\n\n"
# Output: step 5: {'message': 'meet me at the north', 'sender': 7, 'recipients': [1, 2]}

This garbled output was injected into the ReAct reasoning prompt on every step where agents had exchanged messages, silently corrupting the LLM's view of inter-agent communication.

Changes

mesa_llm/memory/memory.py

• Added _format_message_entry(msg_value) -> str helper that renders {"message": "...", "sender": 42, ...} as "Agent 42 says: ...", with a plain-string passthrough for backward compatibility

mesa_llm/memory/episodic_memory.py, st_lt_memory.py, st_memory.py

• Updated get_communication_history() in all three to use _format_message_entry() instead of raw f-string interpolation

Tests

• TestFormatMessageEntry (5 tests) in test_memory_parent.py — unit tests for the helper covering: nested dict with sender, without sender, plain string passthrough, missing message key fallback, and EpisodicMemory's importance-augmented payload
• test_get_communication_history_nested_dict added to all three memory test files — reproduces the real speak_to payload structure and asserts readable output
• test_get_communication_history_skips_non_message_entries added to all three — asserts non-message entries are correctly excluded
• test_get_communication_history_returns_empty_string_when_no_messages added to STLTMemory and ShortTermMemory

Before / After

# Before
step 5: {'message': 'meet me at the north', 'sender': 7, 'recipients': [1, 2]}

# After
step 5: Agent 7 says: meet me at the north

[coderabbitai]

Important

Review skipped

Auto reviews are disabled on this repository. Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Pro

Run ID: 915aeaed-876e-485b-8973-379a3e217672

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 91.27%. Comparing base (27f0f50) to head (88bf4ab).
⚠️ Report is 43 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #194      +/-   ##
==========================================
+ Coverage   90.64%   91.27%   +0.62%     
==========================================
  Files          19       19              
  Lines        1540     1605      +65     
==========================================
+ Hits         1396     1465      +69     
+ Misses        144      140       -4     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[wang-boyu]

Thanks for the fix. I updated the PR so that get_communication_history() keeps the full communication context including recipients:

# Before
step 5: {'message': 'meet me at the north', 'sender': 7, 'recipients': [1, 2]}

# After
step 5: Agent 7 says: meet me at the north

# Now
Step 5: Message from agent 7 to agents 1, 2: meet me at the north

If there's no recipient, then it becomes:

Step 5: Message from agent 7: meet me at the north

Could you pull the latest changes and verify that it's working as expected?