Prompt Codebooks: Discrete Compositional Optimization for Language Model Instruction Refinement
arXiv:2605.28360v1 Announce Type: new Abstract: Automatic prompt optimization (APO) has driven significant gains in LLM-based agentic workflows. However, existing methods treat each task's prompt as a monolithic, instance-blind string optimized through global edits, producing brittle updates and preventing the reuse of learned sub-behaviors. We propose Prompt Codebooks (PCO), a novel compositional prompt optimization framework that recasts APO as discrete learning over a finite vocabulary of natural-language instincts - atomic, reusable instruction units. PCO organizes prompt-construction know
The rapid advancement and adoption of LLM-based agentic workflows highlight the immediate need for more efficient and robust prompt optimization techniques.
This development offers a significant step towards more reliable, reusable, and scalable AI agent construction, improving performance and reducing brittleness in complex AI systems.
Prompt optimization transitions from monolithic, task-specific string edits to a compositional, discrete learning process, enabling reusable 'instincts' for LLMs.
- · AI platform developers
- · Enterprises deploying AI agents
- · Researchers in autonomous AI
- · Open-source AI communities
- · Companies relying on manual prompt engineering
- · Brittle, single-purpose AI agent frameworks
Increased efficiency and reliability in developing and deploying AI agents due to better prompt optimization.
Accelerated development of more complex and general-purpose AI agents capable of handling diverse tasks.
Potential for a 'library economy' of AI instincts, fostering collaboration and standardization in AI agent design.
This signal links to a primary source. Continuum Brief monitors and indexes it as part of the live intelligence stream — we do not republish source content.
Read at arXiv cs.AI