feat: Implement strict validation for client while loading tools

[anubhav756]

This pull request introduces a strict mode to the load_toolset method and refines the validation behavior of the load_tool method.

Benefits

• Ensures tools are configured as intended and are using all provided authentication mechanisms and parameters.
• Helps identify misconfigurations or unused inputs during the tool loading phase, rather than leading to potential runtime issues.
• Makes it explicit when a tool is not interacting with expected configurations.

Changes

load_toolset(strict: bool = False)

The load_toolset method now includes an optional strict flag:

• strict = False (Default Behavior):

  • Validation will fail (raising a ValueError) if at least one authentication token getter (from the auth_token_getters provided) or at least one bound parameter (from the bound_params provided) is not utilized by any of the tools loaded within the toolset. This ensures that all globally provided configurations find a consumer within the set.

• strict = True:

  • Validation becomes significantly more stringent on a per-tool basis. The method will fail if any individual tool loaded within the toolset does not utilize all of the globally provided authentication token getters and all of the globally provided bound parameters.
  • This mode ensures that every tool explicitly engages with every piece of shared configuration passed during the load_toolset call.

load_tool()

• The strict flag has been intentionally omitted from the load_tool method.
• The validation for load_tool is now standardized to be inherently strict: the single loaded tool must utilize all provided auth token getters and all provided bound parameters.
• This decision was made because load_tool processes only one tool at a time. In this context, the nuanced behaviors of strict = True versus strict = False (as defined for load_toolset) do not apply; the expectation for a single tool is always full utilization of its provided configurations.

Implementation

The validation logic leverages the used_auth_keys and used_bound_keys values returned by the internal __parse_tool helper function. This information indicates which specific auth token getters and bound parameters are actively used by each processed tool.

Note

The strict flag available in load_toolset is a runtime parameter that governs the validation behavior during the loading process. It is not persisted as part of the tool's loaded configuration or state.
This is because at the level of an individual tool's direct configuration (e.g., if one were to add an auth token getter directly to a single tool instance), the "strictness" is inherent: that single tool would naturally be expected to use the configuration specifically provided to it. This aligns with the always-strict validation now enforced by the load_tool method.