The Router Within: Eliciting Native Skill Routing from a Frozen LLM
Ruishuo Chen and colleagues at Tsinghua University show that a frozen agent LLM already carries the signal for choosing which skill to load, and build Gavel, a router that reads it out with two trained linear maps and no skill text in the context.
Ask this paper
Problem with current routing: Deployed harnesses preload every skill's metadata into context, which spreads attention and limits library size. Retrieval pipelines move selection out of context but also away from the agent model.
Glance and verdict: The glance projects mid-layer states of the task and of each skill through two linear maps and scores the whole library against per-skill banks built in one forward pass at installation. The verdict resumes the forward passes of shortlisted skills and combines the model's likelihood and yes/no judgment with the glance.
Results: Trained once, Gavel transfers zero-shot to three public benchmarks and to SkillTraj, a new benchmark of 372 simulated trajectories. On Qwen3-32B it beats progressive disclosure and retrieve-and-rerank pipelines that add 1.2B to 16B external parameters by up to 13.4 points, and by up to 21.9 when the need for a skill arises mid-rollout.
Live harness: In a bash-agent harness the same 32B model triggers the correct skill on Skill-Use more often than much larger frontier models running in Codex, and routing accuracy improves as the backbone improves.
Abstract
Skills extend an LLM agent beyond its parametric knowledge, and the gain they promise rests on picking the right one. Deployed harnesses route by preloading every skill's metadata into the context, which disperses the agent's attention and caps the library size. Retrieval pipelines move the selection out of the context, but also out of the agent's capability. We show that the frozen agent LLM already carries the routing signal in its own forward passes, and that two linear maps suffice to read it out with no skill text in the context. Gavel (Glance And Verdict from a frozen LLM) reads it in two steps. A glance projects the task's and each skill's mid-layer states through the two maps, the only parameters trained, and scores the full library against compact per-skill banks that one forward pass builds at installation. A verdict then resumes the shortlisted skills' forward passes and reads the model's own likelihood and yes/no judgment, fused with the glance as a product of experts. Trained once, Gavel transfers zero-shot to three public benchmarks and SkillTraj, our new benchmark of 372 simulated agent trajectories. On Qwen3-32B it outperforms progressive disclosure and retrieve-and-rerank pipelines that add 1.2B to 16B external parameters, by up to 13.4 points on written tasks and up to 21.9 when the need for a skill arises mid-rollout. Routing accuracy improves as the backbone does, and in a bash-agent harness the same 32B triggers the correct skill on Skill-Use more often than far larger frontier models running in Codex.