Apple Machine Learning Research · 10/12/2026, 08:00:00
Self-Compacting Agents: Lightweight Rubrics Enable Autonomous Context Compaction Without Fine-Tuning
To address the loss of critical intermediate results caused by fixed-threshold compaction in long agent trajectories, this paper proposes the Self-Compacting framework. By pairing a compaction tool with a lightweight rubric, it enables models to autonomously decide when to summarize context based on sub-task resolution or trajectory convergence, rather than relying on external triggers. Experiments across six benchmarks and seven models show significant cost reduction per question, revealing a meta-cognitive gap where unprompted models cannot detect context rot, which scaffolds can bridge without training.
SOURCE COVERAGEFull Original Report
Long agent traces composed of chains of thought and tool calls accumulate stale content that anchor subsequent generations, and eventually outgrow the context window. Existing scaffolds mitigate it with fixed-interval compaction triggered at a token threshold. Such triggers pay no heed to trajectory structure, risking discard of partial results mid-derivation or mid-search. We propose SELFCOMPACT, a scaffold that allows the model itself to decide when and how to compact. Specifically, it pairs two inference-time elements: (i) a compaction tool the model invokes to summarize the accumulated context, and (ii) a lightweight rubric specifying when to fire (a sub-task has resolved, or the trajectory is converging) and when to suppress (mid-derivation, or when stuck). Both are needed. The tool alone is unevenly used across open-weight models, often invoked at unhelpful moments or not at all; the rubric alone cannot act. Together, they elicit effective adaptive compaction without any fine-tuning or external supervision. We present empirical results on six benchmarks (competitive math and agentic search) and seven models. Our results show that SELFCOMPACT matches or exceeds fixed-interval summarization at a fraction of the token cost, improving over a no-summarization baseline by up to 18.1 points on math and 5–9 points on agentic search at 30–70% lower per-question cost. Our results expose a meta-cognitive gap: although unprompted models cannot reliably tell when their own context is rotting, a lightweight rubric closes this gap, reframing when to compact as a capability that scaffolds can supply without training.
- † Johns Hopkins University