browser_run

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

Annotations only say readOnlyHint=false and openWorldHint=true, yet the description adds substantial behavioral context: fail-fast ordering, partial application of earlier steps on failure, which step is named in the error, an implicit final snapshot, the WebMCP discovery/call protocol, the confirm requirement for non-read-only page tools, and the explicit warning that page tool output is untrusted. That is exactly the mutation and trust-boundary disclosure annotations cannot provide.

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?

Front-loaded with the execution and failure contract, then the step grammar, then guidance, then the WebMCP addendum — a sensible ordering. The WebMCP sentences are dense and slightly tangled with nested parentheticals, but no sentence is pure filler given the untyped steps array.

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?

For a mutating, open-world batch executor this covers ordering, failure behavior, partial state, step grammar, and trust of external output. With no output schema, the return contract is only partly specified — it says a snapshot is appended but not how a failed run reports the partial results or the error shape — which leaves a modest gap.

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?

The schema documents coverage at 100% but the steps array is a bare untyped object with no enums, so the description carries the entire operation vocabulary ('click'|'fill'|'type'|'press'|'select'|'hover'|'scroll'|'wait'|'goto'|'eval') plus the fields each op expects and the two WebMCP pseudo-ops. Without this text the array would be unusable; the description is doing real semantic work beyond the schema.

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 opens with a precise verb and resource ('run several steps in order'), states the failure contract, and implicitly contrasts with the single-operation siblings (browser_click, browser_type, browser_fill, etc.) by being the batching entry point. An agent can tell immediately that this is the multi-step executor rather than a one-off action.

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?

'Chain steps whose outcome you are sure of; look before acting on anything uncertain' gives a clear rule for when to batch versus when to take a snapshot first. It also scopes the WebMCP confirm flag ('only for a tool not marked read-only, and only when the user asked for what it does'). It stops short of naming the specific sibling to fall back on by name, so it is strong but not fully explicit routing.

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