fix: add serialization-safe agent_id and to_dict() to MemoryEntry (#156)

[aatmaj28]

Summary
Fixes #156 — MemoryEntry stores a direct LLMAgent object reference in its agent field, which causes TypeError: Object of type LLMAgent is not JSON serializable when memory entries are serialized.

Root Cause
The MemoryEntry dataclass (in mesa_llm/memory/memory.py) has agent: "LLMAgent" as a field. Every memory module (st_memory, lt_memory, st_lt_memory, episodic_memory) creates entries with agent=self.agent, embedding the full Agent object. Any downstream code that attempts json.dumps() on a MemoryEntry or its fields fails.

Changes

• Added agent_id: int | None = None field to MemoryEntry with post_init that auto-populates from agent.unique_id
• Added to_dict() method that returns a JSON-serializable dictionary (content, step, agent_id)
• Fully backwards compatible: the existing agent field is unchanged, so all 9 call sites across 4 memory modules continue to work without modification

Testing
Existing tests unaffected — change is purely additive (new field with default + new method).

Context
I'm a GSoC 2026 candidate working on the Mesa-LLM stabilization project proposal. This is one of the critical issues I identified for Phase 1.

[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: fa1791c6-7c7d-4f29-9ded-35a269b72fd7

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.}

Tip

CodeRabbit can use your project's `ruff` configuration to improve the quality of Python code reviews.

Add a Ruff configuration file to your project to customize how CodeRabbit runs ruff.

[IlamaranMagesh]

Coul you please elaborate what's the issue here? I see the issue #156 is already closed and it has some fixes on JSON serialisation error.

[aatmaj28]

Coul you please elaborate what's the issue here? I see the issue #156 is already closed and it has some fixes on JSON serialisation error.

Hi @IlamaranMagesh , thanks for the review! You're right that #156 was closed, but looking at the fix that was merged, it addressed the send_message/asend_message methods specifically.
My PR addresses a related but different aspect: the MemoryEntry dataclass itself still stores the full LLMAgent object in its agent field (line 22 of memory.py). While the content dictionaries passed to add_to_memory were fixed to use unique_id, the MemoryEntry object that wraps them still holds a non-serializable agent reference. This means any code that tries to serialize a MemoryEntry directly (e.g., for logging, export, or passing to an LLM context) will still fail.
The agent_id field and to_dict() method I added provide a clean serialization path without breaking the existing agent field that other code depends on. Happy to close this if the maintainers feel it's not needed though!

[wang-boyu]

Closing because #156 was already fixed, and this PR does not address that code path. Making MemoryEntry json-serializable could be a useful enhancement, but there is no current built-in logic that requires it. If that need comes up, we can revisit it in the future. Thanks for the effort though.