Multi-round reflection lets agents built on large language models recover from failures within a single task, but lessons learned across many reflection rounds on one task are discarded before the next begins. We study this gap on a code-generation task: from a scientific paper section, the agent writes executable Manim code. We present ManimAgent, a self-evolving multimodal agent that carries reflection experience across tasks through a dual-channel Episodic Memory Bank grown entirely from its own task stream, with no weight updates and no human seeds. After each animation converges, a vision–language model scores the rendered keyframes; the resulting signals populate a positive channel ℳ⁺ that stores success rationales as soft Reference Examples, and a negative channel ℳ^- that stores VLM-screened failure patterns as hard Known Pitfalls. On a fixed-probe evaluation against no-memory, matched-budget retrieval-augmented generation, and shuffled-memory baselines, blind human Pass@1 rises and reflection rounds fall as memory size grows. We will release the code, frozen memory snapshots, and the task stream.