ui_refine

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

With no annotations provided, the description carries the full burden of behavioral disclosure. It clearly describes the destructive/transformative operations (deleting, merging, labeling, adding text anchors) and what the tool does NOT do (coordinate fine-tuning). It also discloses the return value (corrected element list and change log). However, it does not discuss side effects on input files, failure modes, or prerequisites beyond the image, so it is not fully transparent.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Is the description appropriately sized, front-loaded, and free of redundancy?

The description is two sentences with no filler. It front-loads the main purpose, then lists operations, then provides an exclusion, and ends with the return type. Every clause earns its place, making it highly efficient and well-structured.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

The description explains the tool's purpose, operations, and return value, but fails to clarify a critical aspect: how are the UI detection boxes provided as input? The schema only includes an image parameter, yet the tool is supposed to correct detection boxes. This is a significant gap that leaves the agent uncertain whether boxes are drawn in the image, passed via internal state, or expected to be detected by the VLM itself. Given the moderate complexity and lack of output schema, the description should have addressed this.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema description coverage is 67% (coords and out_path have descriptions, image does not). The tool description does not add any parameter-specific meaning beyond the schema. The baseline for this coverage is 3, and the description does not improve it, so a score of 3 is appropriate.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Does the description clearly state what the tool does and how it differs from similar tools?

The description clearly states the tool's function with a specific verb: 'VLM 审查并语义修正 UI 检测框' (reviews and semantically corrects UI detection boxes). It lists concrete operations (delete false positives, merge duplicates, semantic labeling, text anchor gap-filling) and explicitly distinguishes itself from coordinate-level correction, which is handled by som/cv. This differentiates it clearly from sibling tools like cv_locate and som_locate.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Does the description explain when to use this tool, when not to, or what alternatives exist?

The description provides an explicit exclusion: '不做坐标微调（坐标级修正由 som/cv 负责）' (does not do coordinate fine-tuning, coordinate-level correction is handled by som/cv), naming the alternative tool for that use case. It implies usage after initial detection to clean up false positives and duplicates, but does not fully enumerate all possible alternatives like ui_parse or ui_locate.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.