Latent Briefing

muratcankoylan/agent-skills-for-context-engineering

If you're running orchestrator-worker agent systems and watching token costs compound as the orchestrator's trajectory gets replayed to every worker, this skill shows you how to share state at the KV cache level instead of text. It adapts Attention Matching compaction with task-guided queries and robust thresholding to keep only the positions the worker actually needs for its current subtask. The technique requires runtime access to KV tensors, so it won't work through standard APIs, but the documented results show material worker-token savings on long-document QA with low-single-digit-second overhead. It's a specialized primitive for systems where summarization is too lossy and full context replay is too expensive.

skills.sh installs: 195
GitHub stars: 18.3K
last updated: May 26, 2026

View on GitHub

Install

$ npx -y skills add muratcankoylan/agent-skills-for-context-engineering --skill latent-briefing --agent claude-code

Installs into .claude/skills of the current project.

Files

Latent Briefing and KV Cache Memory Sharing

Hierarchical multi-agent systems often pay for the same context twice. The orchestrator accumulates a long reasoning trajectory, but each worker usually receives only a narrow text handoff such as a subtask prompt plus raw document slices. Passing the full trajectory fixes coverage but drives token cost up on every worker call. Summarization introduces latency and information loss. Retrieval helps with document access but does not preserve the orchestrator's evolving reasoning state.

Latent Briefing addresses this by sharing memory at the representation level rather than the text level. The core idea is to compact the orchestrator trajectory in the worker model's KV cache, keeping positions that are most relevant to the current worker task. The method builds on Attention Matching (AM) KV cache compaction and adapts it for inference-time multi-agent handoff with task-guided queries, a shared token mask across heads, and robust thresholding.