Compression Remembers the Detour: How Context Compression Amplifies Off-Task Content and Derails LLM Agents
Long-horizon LLM agents rely on context compression to keep growing histories within a finite window [1][2]. These mechanisms are goal-agnostic—retention follows volume, recurrence, and generic salience rather than task relevance—and we show they systematically amplify off-task content, inducing Compression-Induced Goal Drift (CIGD). We formalize CIGD via the drift amplification ratio (DAR), evaluate six compression strategies on a controlled injection benchmark ( per cell), and propose Goal-Anchored Compression (GAC): a pinned goal anchor, a negation ledger, status-aware retention scoring, and drift re-anchoring. Volume- and recurrence-based compressors amplify detours monotonically with repetition (attention-scored eviction reaches at twelve-fold repetition), while query-conditioned compressors de-amplify. Under severe budgets, summarization and scored eviction lose the user’s negated constraint in and of episodes where GAC retains it in ; GAC eliminates residual detour content () with smaller contexts. Prior work asks what compression forgets [3][4]; we ask what it amplifies.