AGORA: Adapter-Grounded Observation-Action Retention for Inference-Free Prompt Compression in LLM Agents

In three linesAGORA introduces inference-free prompt compression for LLM agents using adapter-grounded observation-action retention. Standard token-level extractive compressors fail on agents (75% performance in 8/9 cases). Ablation shows structure and adaptive scorer enable 1.0-11.5x compression from fixed keep ratio.

Summary generated by Claude — human-verified