Capability-Grounded Agentic Planning

A new arXiv paper (2610.41040v1) introduces Capability-Grounded Agentic Planning (CGAP), a method that compiles a tool catalog deterministically into a PDDL model, uses an LLM to convert requests into goals, and solves the model with a classical planner rather than trial-and-error tool selection. Tested with three LLMs across ToolSandbox, ComplexFuncBench, BFCL multi-turn, and τ²-bench retail, CGAP outperforms a standard tool-calling agent by up to 28.9 points. The approach tracks which inputs are available rather than their values, adds entity-specific calls during execution, and requires an LLM verifier to confirm each requested result before counting it as delivered.

arXiv:2610.41040v1 Announce Type: cross Abstract: LLM agents serve natural-language requests through tools, data sources, and other agents, and even simple requests often need several interdependent calls. Typically, the LLM picks these calls by trial and error, with no guarantee of an executable plan. We present Capability-Grounded Agentic Planning CGAP , which instead reasons about what a request requires and what tools provide. CGAP compiles the tool catalog deterministically into a PDDL model, uses an LLM to turn the request into goals, and solves the model with a classical planner. Since tool outputs are unknown in advance, the model is abstract, tracking which inputs are available, not their values. During execution, an LLM adds entity-specific calls that abstract plans omit, and deterministic checks reject any that break the plan. Every argument is drawn from a cited source, such as a conversation span or an earlier tool response, and each requested result counts as delivered only after an LLM verifier checks it against the request. With three LLMs on ToolSandbox, ComplexFuncBench, BFCL multi-turn, and $\tau^2$-bench retail, CGAP outperforms a standard tool-calling agent by up to 28.9 points.