Progressive Skill Discovery as Access Control for Tool-Using LLM Agents: Structural Governance through Role-Scoped Capability Delivery

arXiv:2609.28693v1 Announce Type: new Abstract: Large Language Model (LLM) agents struggle to scale safely when exposed to vast enterprise toolsets. Providing an agent with access to every internal tool leads to oversized context windows, degraded tool selection, and severe governance vulnerabilities - as system policies defined purely in prompts remain probabilistic advice rather than hard constraints. Existing mitigations, such as multi-agent domain delegation, decentralize audit logs and fail to guarantee policy compliance across sessions. We introduce skilder, a framework that packages capabilities into roles: bundles of skills, tools, and instructions, together with the limits that bound them. An agent begins with a minimal role catalog, learns the roles a task requires, and receives each role's skills, instructions, and tools through a single MCP server. Because tools reach the agent only inside learned skills, the same server enforces the scope of what was learned deterministically. We evaluate skilder against flat-context tool selection and multi-agent orchestration across 13 tasks using six models (10 runs each). Our results show that, when models completed discovery and issued a governed call, the skilder simulated authorization layer enforced governance boundaries: no unauthorized tool call or parameter violation (e.g., a spending-limit breach) executed. Aggregate task pass rates also reflect whether each model followed the discovery protocol and satisfied response-quality checks; those misses are not authorization failures. Furthermore, by allowing agents to dyn

skilder framework provides deterministic governance boundaries

The framework skilder introduces deterministic governance for LLM agents' tool usage, reducing unauthorized actions and improving security. This enhances safety in AI tool integration but could limit flexibility or introduce new governance dependencies. The evidence is experimental but focused on compliance in controlled settings, suggesting moderate positive impact with some potential governance trade-offs.

The skilder framework aims to improve the safety and governance of LLM agents by providing deterministic enforcement of access policies for enterprise toolsets. This technical advancement reduces the risk of unauthorized actions and data breaches by limiting tool access, which is a positive implication for the security and reliability of AI systems.

SOURCE     arXiv cs.AI (tier 1)
   ↓
DOCUMENT   2fab7939-add2-4bb9-8616-8756a0dbc2ba
           https://arxiv.org/abs/2609.28693
   ↓
EVIDENCE   3 extracted excerpts
   ↓
MODEL RUN  4 runs, methodology 1.0.1
   ↓
SCORE      +2  (Mixed or uncertain)
   ↓
CONFIDENCE 73%