Add more custom properties and checks throughout

[egenn]

No description provided.

[gemini-code-assist]

Code Review

This pull request introduces several S7 property helpers and validation functions, including enum, check_logical_scalar, and check_pos_integer_scalar. It refactors the Field, ToolParameter, and Agent classes to utilize these new helpers, which simplifies their internal validation logic. Additionally, the PR implements a uniqueness check for tool parameter names and updates the available_tools function to use message() for output formatting. Review feedback identifies a missing definition for the check_integer_scalar function, suggests adding explicit NA handling to the enum validator, and recommends standardizing the naming of logical scalar check functions across the package.

[gemini-code-assist]

The function check_integer_scalar is called here but it is not defined in this file or anywhere else in the provided codebase. This will result in a "could not find function" error at runtime. Please ensure this helper is defined or correctly imported.

[Copilot]

Pull request overview

This PR expands internal S7 property helpers and validation to tighten schema/tool/agent correctness across the package.

Changes:

• Introduces an enum() S7 property helper and new check helpers, and uses them in Field / ToolParameter.
• Adds additional validation (e.g., unique tool parameter names; new create_agent() argument checks).
• Adjusts available_tools() output behavior (stdout → stderr) and tweaks an Agent duplicate-tool error message.

Reviewed changes

Copilot reviewed 4 out of 4 changed files in this pull request and generated 5 comments.

File	Description
r/R/00_init.R	Adds enum() + new check helpers; changes available_tools() to print via message()
r/R/01_Schema.R	Refactors Field to use enum(.SCHEMA_FIELD_TYPES) and simplifies validation
r/R/04_Tool.R	Refactors ToolParameter to use enum() and adds duplicate parameter-name validation in Tool
r/R/07_Agent.R	Adds argument checks in create_agent() and adjusts duplicate tool-name validator messaging

Comments suppressed due to low confidence (1)

r/R/00_init.R:858

• available_tools() switched from cat() (stdout) to message() (stderr). This is a user-visible behavior change and can break callers that capture/pipe stdout (e.g., capture.output()) while also making the output look like diagnostics. Consider reverting to stdout (cat()/cli::cat_line()/cli::cli_inform() depending on the intended UX) so the function remains a pure “print info” helper.

}

# %% available_tools ----
#' Print built-in tools available for use by agents
#'
#' Prints the R handle (`tool_*`), the `function_name` the model sees, and the
#' description of every built-in `Tool` exported by the package. Derived at
#' call time from the namespace — no hardcoded list.
#'
#' @param verbosity Integer: Verbosity level.
#'
#' @return A named list of `Tool` objects keyed by their R handle, invisibly.
#'

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

New validation rejects duplicate tool parameter names, but there’s no test covering this behavior (and it’s easy to regress). Add a unit test that constructs a Tool with two parameters sharing the same name and asserts the expected error.

[Copilot]

create_agent() now validates max_tool_rounds via check_pos_integer_scalar(), but there’s no test asserting that invalid values (0, negative, non-integer) are rejected (or that valid values are accepted). Adding a small test around this new argument validation will help prevent accidental loosening/tightening of the API contract.