Agent Engineering Wiki

"Context engineering and memory management" has emerged as a discipline of its own — treating the prompt as a managed working set rather than an append-only log. Techniques range from rolling summarization to LLM-guided compression of long-term memory (MemRefine) and memory systems that explicitly model association, forgetting, and synthesis rather than storing everything. Compaction is increasingly paired with an external store: compress the working set, offload the rest to a vector/graph KB, and rehydrate on demand. A complementary, cheaper move is compaction at the input boundary — shrinking a tool result *before* it ever enters the context, not summarizing it afterward. Coding agents read verbose build/test logs, so deterministic pre-compactors that strip noise from that output (Logslim) cut the per-step token bill with no model call and no lossy summarization of the agent's own reasoning. The newest finding is that compaction is not just lossy but safety-critical: "Governance Decay" shows that summarizing, evicting, or compressing context in a long-horizon agent can silently drop the very safety/governance constraints that were stated up front, so a later step acts as if rules it was given hours ago no longer apply — the compactor is a security surface, not just a cost optimization.

Compaction now has a documented safety failure mode: "Governance Decay" shows that context summarization/eviction in long-running agents can silently erase the safety and governance constraints set earlier in the session, reframing the compactor as a security-critical layer that needs constraint-preserving guarantees — not just a token-saving one. That sits alongside smarter compression (LLM-guided MemRefine, forgetting/synthesis-aware stores) and input-boundary trimming of verbose tool output (Logslim).

Cheap on infra (no external store) and keeps everything the model needs in one place, but summarization is lossy and irreversible — a detail dropped early can't be recovered later, and aggressive compaction can quietly degrade task fidelity. Best for single-session, long-horizon tasks where recency dominates and the full history isn't needed verbatim. The sharpest failure mode is not lost task detail but lost *constraints*: Governance Decay shows compaction can quietly evict the safety/policy rules an agent was given up front, so over a long session it drifts out of its guardrails — which means anything load-bearing (permissions, safety limits, the user's hard "do not") must be pinned outside the compactible window, not left to survive summarization (see prompt injection).

Often the highest-leverage first move: it directly attacks token cost and latency (the bill scales with context size) without standing up new infrastructure. The risk is silent quality loss, so it needs evaluation — which makes it a tuning knob, not a set-and-forget fix.