context-optimization

# Context Optimization Techniques Context optimization extends the effective capacity of limited context windows through compression, masking, caching, and partitioning, applied with measurement discipline. The techniques below are ordered by impact and risk. ## When to Activate - Context budgets or token costs constrain task complexity - Observation masking can replace verbose tool outputs with retrievable references - Prefix or KV-cache hit rate needs improvement - Retrieval scoping can reduce irrelevant loaded context - Context partitioning can extend effective capacity across agents - Budget triggers are needed for masking, compaction, or partitioning - Agent quality drops as context grows, or output is wrong or irrelevant for no obvious reason (diagnose mode) ## Modes | Mode | Use when | Read | |------|----------|------| | Optimize (default) | Token cost or budget limits the task and the context is behaving | This file, then [references/optimization_techniques.md](./references/optimization_techniques.md) | | Diagnose | Quality has already dropped: lost-in-middle, poisoning, distraction, confusion, or clash | [references/degradation.md](./references/degradation.md), then [references/degradation-patterns.md](./references/degradation-patterns.md) for detection code and recovery | | Explain | Onboarding to context mechanics with no active failure | [references/fundamentals.md](./references/fundamentals.md) | Diagnose before optimizing a failing agent: compaction or masking applied to a poisoned or clashing context carries the error forward. First confirm the prompt works at low context length (if it fails at 2K tokens, the prompt is the problem), then match the symptom to a pattern and apply its first move: | Pattern | Typical symptom | First move | Optimization tactic | |---------|-----------------|------------|---------------------| | Lost-in-middle | Correct fact in context is ignored; early instructions forgotten | Move critical content to the start or end; prepend a summary | Compaction that keeps key findings at the edges | | Poisoning | A wrong claim persists after correction; tools misused | Truncate to before the bad claim and reload verified sources only | Restart with a clean partition | | Distraction | Quality falls after loading extra documents | Filter retrieval before loading; put reference material behind tool calls | Retrieval scoping, observation masking | | Confusion | Constraints or tools from another task leak in | Split tasks into separate contexts | Context partitioning | | Clash | Contradictory sources give unpredictable answers | Set source precedence and drop outdated versions before loading | Retrieval scoping | Trigger compaction well before the model-specific onset of degradation (about 70% of the known onset), not after symptoms appear. ## Core Concepts Apply four primary strategies in this priority order: 1. **KV-cache optimization** — Reorder and stabilize prompt structure so the inference engine reuses cached Key/Value tensors. This is the cheapest optimization when the runtime supports prefix caching: low quality risk, immediate cost and latency savings. Apply it first when stable prefixes exist. 2. **Observation masking** — Replace verbose tool outputs with compact references once their purpose has been served. Tool outputs can dominate agent trajectories (claim-context-optimization-tool-output-dominance), so masking often yields the largest capacity gains. The original content remains retrievable if needed downstream. 3. **Compaction** — Summarize accumulated context at about 70% of the model's measured degradation onset (not 70% of the full window), then reinitialize with the summary. This distills the window's contents while preserving task-critical state. Compaction is lossy — apply it after masking has already removed the low-value bulk. 4. **Context partitioning** — Split work across sub-agents with isolated contexts when a single window cannot hold the full problem. Each sub-agent operates in a clean context focused on its subtask. Reserve this for tasks where estimated context exceeds 60% of the window limit, because coordination overhead is real. The governing principle: context quality matters more than quantity. Every optimization preserves signal while reducing noise. Measure before optimizing, then measure the optimization's effect. ## Detailed Topics ### Compaction Strategies Trigger compaction at about 70% of the measured degradation onset (with onset at 60% of the window, that is about 42% utilization): summarize the current context, then reinitialize with the summary. This distills the window's contents in a high-fidelity manner, enabling continuation with minimal performance degradation. Prioritize compressing tool outputs first (they consume 80%+ of tokens), then old conversation turns, then retrieved documents. Never compress the system prompt — it anchors model behavior and its removal causes unpredictable degradation. Preserve different elements by message type: - **Tool outputs**: Extract key findings, metrics, error codes, and conclusions. Strip verbose raw output, stack traces (unless debugging is ongoing), and boilerplate headers. - **Conversational turns**: Retain decisions, commitments, user preferences, and context shifts. Remove filler, pleasantries, and exploratory back-and-forth that led to a conclusion already captured. - **Retrieved documents**: Keep claims, facts, and data points relevant to the active task. Remove supporting evidence and elaboration that served a one-time reasoning purpose. Target 50-70% token reduction with less than 5% quality degradation. If compaction exceeds 70% reduction, audit the summary for critical information loss — over-aggressive compaction is the most common failure mode. ### Observation Masking Mask observations selectively based on recency and ongoing relevance — not uniformly. Apply these rules: - **Never mask**: Observations critical to the current task, observations from the most recent turn, observations used in active reasoning chains, and error outputs when debugging is in progress. - **Mask after 3+ turns**: Verbose outputs whose key points have already been extracted into the conversation flow. Replace with a compact reference: `[Obs:{ref_id} elided. Key: {summary}. Full content retrievable.]` - **Always mask immediately**: Repeated/duplicate outputs, boilerplate headers and footers, outputs already summarized earlier in the conversation. Masking should achieve 60-80% reduction in masked observations with less than 2% quality impact. The key is maintaining retrievability — store the full content externally and keep the reference ID in context so the agent can request the original if needed. ### KV-Cache Optimization Maximize prefix cache hits by structuring prompts so that stable content occupies the prefix and dynamic content appears at the end. KV-cache stores Key and Value tensors computed during inference; when consecutive requests share an identical prefix, the cached tensors are reused, saving both cost and latency. Apply this ordering in every prompt: 1. System prompt (most stable — never changes within a session) 2. Tool definitions (stable across requests) 3. Frequently reused templates and few-shot examples 4. Conversation history (grows but shares prefix with prior turns) 5. Current query and dynamic content (least stable — always last) Design prompts for cache stability: remove timestamps, session counters, and request IDs from the system prompt. Move dynamic metadata into a separate user message or tool result where it does not break the prefix. Even a single whitespace change in the prefix invalidates the entire cached block downstream of that change. Compaction or masking of earlier history also requires processing the following text again; batch these edits and prefer edits near the end when the runtime allows it. Measure the cache and latency impact rather than assuming token reduction is free. Target 70%+ cache hit rate for stable workloads. At scale, this translates to 50%+ cost reduction and 40%+ latency reduction on cached tokens. ### Context Partitioning Partition work across sub-agents when a single context cannot hold the full problem without triggering aggressive compaction. Each sub-agent operates in a clean, focused context for its subtask, then returns a structured result to a coordinator agent. Plan partitioning when estimated task context exceeds 60% of the window limit. Decompose the task into independent subtasks, assign each to a sub-agent, and aggregate results. Validate that all partitions completed before merging, merge compatible results, and apply summarization if the aggregated output still exceeds budget. This approach achieves separation of concerns — detailed search context stays isolated within sub-agents while the coordinator focuses on synthesis. However, coordination has real token cost: the coordinator prompt, result aggregation, and error handling all consume tokens. Only partition when the savings exceed this overhead. ### Budget Management Allocate explicit token budgets across context categories before the session begins: system prompt, tool definitions, retrieved documents, message history, tool outputs, and a reserved buffer (5-10% of total). Monitor usage against budget continuously and trigger optimization when any category exceeds its allocation or total utilization crosses 70%. Use trigger-based optimization rather than periodic optimization. Monitor these signals: - Token utilization above 80% — trigger compaction - Attention degradation indicators (repetition, missed instructions) — trigger masking + compaction - Quality score drops below baseline — audit context composition before optimizing ## Practical Guidance ### Optimization Decision Framework Select the optimization technique based on what dominates the context: | Context Composition | First Action | Second Action | |---|---|---| | Tool outputs dominate (>50%) | Observation masking | Compaction of remaining turns | | Retrieved documents dominate | Summarization | Partitioning if docs are independent | | Message history dominates | Compaction with selective preservation | Partitioning for new subtasks | | Multiple components contribute | KV-cache optimization first, then layer masking + compaction | — | | Near-limit with active debugging | Mask resolved tool outputs only — preserve error details | — | ### Performance Targets Track these metrics to validate optimization effectiveness: - **Compaction**: 50-70% token reduction, <5% quality degradation, <10% latency overhead from the compaction step itself - **Masking**: 60-80% reduction in masked observations, <2% quality impact, near-zero latency overhead - **Cache optimization**: 70%+ hit rate for stable workloads, 50%+ cost reduction, 40%+ latency reduction - **Partitioning**: Net token savings after accounting for coordinator overhead; break-even typically requires 3+ subtasks Iterate on strategies based on measured results. If an optimization technique does not measurably improve the target metric, remove it — optimization machinery itself consumes tokens and adds latency. ## Examples **Example 1: Compaction Trigger** ```python if context_tokens / context_limit > 0.8: context = compact_context(context) ``` **Example 2: Observation Masking** ```python if len(observation) > max_length: ref_id = store_observation(observation) return f"[Obs:{ref_id} elided. Key: {extract_key(observation)}]" ``` **Example 3: Cache-Friendly Ordering** ```python # Stable content first context = [system_prompt, tool_definitions] # Cacheable context += [reused_templates] # Reusable context += [unique_content] # Unique ``` **Example 4: Budget-triggered optimization policy** ```yaml budgets: tool_outputs: 35% message_history: 30% retrieved_documents: 20% reserved_buffer: 15% triggers: tool_outputs_over_budget: mask resolved observations total_context_over_70_percent: compact message history repeated_irrelevant_retrievals: tighten retrieval scope ``` ## Guidelines 1. Measure before optimizing—know your current state 2. Apply masking before compaction — remove low-value bulk first, then summarize what remains 3. Design for cache stability with consistent prompts 4. Partition before context becomes problematic 5. Monitor optimization effectiveness over time 6. Balance token savings against quality preservation 7. Test optimization at production scale 8. Implement graceful degradation for edge cases ## Gotchas 1. **Whitespace breaks KV-cache**: Even a single whitespace or newline change in the prompt prefix invalidates the entire KV-cache block downstream of that point. Pin system prompts as immutable strings — do not interpolate timestamps, version numbers, or session IDs into them. Diff prompt templates byte-for-byte between deployments. 2. **Timestamps in system prompts destroy cache hit rates**: Including `Current date: {today}` or similar dynamic content in the system prompt forces a full cache miss on every new day (or every request, if using time-of-day). Move dynamic metadata into a user message or a separate tool result appended after the stable prefix. 3. **Compaction under pressure loses critical state**: When the model performing compaction is itself under context pressure (>85% utilization), its summarization quality degrades — it omits task goals, drops user constraints, and flattens nuanced state. Trigger compaction at 70-80%, not 90%+. If compaction must happen late, use a separate model call with a clean context containing only the material to summarize. 4. **Masking error outputs breaks debugging loops**: Over-aggressive masking hides error messages, stack traces, and failure details that the agent needs in subsequent turns to diagnose and fix issues. During active debugging (error in the last 3 turns), suspend masking for all error-related observations until the issue is resolved. 5. **Partitioning overhead can exceed savings**: Each sub-agent requires its own system prompt, tool definitions, and coordination messages. For tasks with fewer than 3 independent subtasks, the coordination overhead often exceeds the context savings. Estimate total tokens (coordinator + all sub-agents) before committing to partitioning. 6. **Cache miss cost spikes after deployment changes**: Reordering tools, rewording the system prompt, or changing few-shot examples between deployments invalidates the entire prefix cache, causing a temporary cost spike of 2-5x until the new cache warms up. Roll out prompt changes gradually and monitor cache hit rate during deployment windows. 7. **Compaction creates false confidence in stale summaries**: Once context is compacted, the summary looks authoritative but may reflect outdated state. If the task has evolved since compaction (new user requirements, corrected assumptions), the summary silently carries forward stale information. After compaction, re-validate the summary against the current task goal before proceeding. ## Integration This skill owns token-efficiency tactics, budget policy, and degradation diagnosis. Adjacent skills own storage and architecture: - `multi-agent-patterns`: partitioning work across isolated agent contexts. - `evaluation`: measuring whether the optimization improved quality, cost, or latency. - `memory-systems`: persistent retrieval layers that feed context just in time. ## References Internal references: - [Context Fundamentals](./references/fundamentals.md) - Read when: explaining what context is, window anatomy, attention mechanics and the U-shaped curve, or onboarding someone to context engineering; also carries the component-by-component reference and `scripts/context_manager.py` usage - [Optimization Techniques Reference](./references/optimization_techniques.md) - Read when: implementing a specific optimization technique and needing detailed code patterns, threshold tables, or integration examples beyond what the skill body provides - [Context Degradation](./references/degradation.md) - Read when: diagnosing why agent performance has dropped and needing to identify which degradation pattern is occurring before selecting an optimization; [degradation-patterns.md](./references/degradation-patterns.md) holds the detection code and recovery procedures, and `scripts/degradation_detector.py` runs the checks Related skills in this collection: - evaluation - Read when: setting up metrics and benchmarks to measure whether an optimization technique actually improved outcomes External resources: - Research on context window limitations - Read when: evaluating model-specific context behavior (e.g., lost-in-the-middle effects, attention decay curves) - KV-cache optimization techniques - Read when: implementing prefix caching at the inference infrastructure level (vLLM, TGI, or cloud provider APIs) - Production engineering guides - Read when: deploying context optimization in a production pipeline and needing operability patterns (monitoring, alerting, rollback)