browser-use

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

With no annotations, the description carries the full burden, and it does meaningful work: it discloses the polling pattern and enumerates the return contents (status values, step count, output, cost breakdown, live URL). It omits auth requirements, rate limits, and error behavior for an invalid session_id, which keeps it below 5.

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?

Three short sentences with zero filler, front-loaded with what the tool does, then usage, then return shape. Every sentence earns its place.

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 single-parameter read tool with no output schema, the description supplies purpose, polling guidance, and a full enumeration of returned fields, so an agent has everything needed to call and interpret it. Nothing material is missing.

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 100% and the single session_id parameter is documented in the schema as a UUID. The description adds no parameter-level detail beyond that, so the baseline 3 for high-coverage schemas applies.

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?

States a specific verb+resource (get a session's status and output) and scopes it precisely to current state, which separates it from list_sessions and get_session_messages. It stops short of naming those siblings or contrasting with them explicitly, so an agent must infer the boundary.

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?

Gives clear usage context: 'Call this repeatedly to poll for task completion,' which tells the agent exactly when and how to invoke it. It does not state when-not to use it or point to stop_session/run_session as alternatives, so no exclusions are covered.

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

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

With no annotations, the description carries the full burden. It discloses the content of the returned messages (browser actions, reasoning, results) and the pagination behavior, which is genuine value, but it omits read-only semantics, ordering guarantees, and any auth or rate-limit context.

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?

Three tight sentences with the purpose front-loaded and the pagination hint last; nothing is redundant. It stops just short of being maximally efficient.

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 simple read tool with no annotations and no output schema, the description covers the essential ground: what is returned and how to page through it. It could say more about message ordering or the response shape, but the core agent needs are met.

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 coverage is 100%, so all three parameters are already documented in the schema, making 3 the baseline. The description reinforces the cursor-based polling use of after= but adds no syntax or format detail beyond what the schema states.

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 states a specific verb+resource (get agent messages/steps for a session) and adds substance by naming what the messages contain: browser actions, reasoning, results. It does not explicitly differentiate itself from the sibling get_session, so it falls short of a 5.

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?

It gives clear operational context: use the after= parameter to poll for new messages since your last check. That tells the agent the intended polling workflow, but there are no stated exclusions or explicit comparisons to alternatives like get_session.

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

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

No annotations are provided, so the description carries the full behavioral burden. It adds useful context that profiles store persistent authentication (cookies, sessions) for login-requiring sites, but it does not disclose return format details, pagination, or permission requirements.

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?

Three tight sentences, front-loaded with the action and scope, followed by the resource concept and then the recommended workflow. Every sentence earns its place with no redundancy.

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 simple list tool with no input parameters and no output schema, the description is mostly complete: it explains purpose, usage context, and reveals the key output field (profile_id) needed for run_session. It could still describe the profile list structure more explicitly.

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 tool takes zero parameters, so there is no parameter schema to compensate for. Per the baseline rule for 0 parameters, a 4 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?

States a specific verb ('List') and resource ('cloud browser profiles') scoped to the authenticated project. The sibling tools are all session-related, and this tool's resource is clearly distinct, so an agent can identify it without confusion.

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?

Explicitly says to use it 'BEFORE attempting any authenticated tasks' and gives a workflow: show profiles to the user, let them choose, then pass the selected profile_id to run_session. It lacks explicit when-not guidance or alternatives for skipping the list step, but the context is strong.

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

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

With no annotations, the description carries the entire behavioral burden. It does disclose that results include status and cost info, but says nothing about ordering ('recent' is undefined), how pagination behaves, default result volume, or any auth/permission requirements for the authenticated project.

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?

A single well-formed sentence with no filler and the key scope ('authenticated project') front-loaded. Its brevity edges into under-specification, which is captured in other dimensions rather than penalized here.

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 read-only list tool with no annotations and no output schema, the description covers purpose and partial return content but omits ordering behavior and pagination semantics. It is minimally adequate rather than complete.

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 100% (page and page_size are both documented with defaults and bounds), so the baseline is 3. The description adds no parameter meaning beyond what the schema already provides.

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 states a specific verb and resource ('List recent sessions') and scopes them to 'the authenticated project', plus it previews the payload ('status and cost info'). It does not explicitly distinguish itself from sibling get_session, so it falls short of the top mark.

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?

Usage is only implied: 'List recent sessions' naturally contrasts with the singular get_session, but no when-to-use, when-not-to-use, or alternative (e.g. get_session for a single session) is named. An agent can infer the listing intent but gets no routing guidance.

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

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

With no annotations, the description carries the full burden and does disclose key behavior: this is the v3 sandbox-based agent, sessions are long-running, and keep-alive/workspaces are supported. The polling model (async task execution, not a synchronous result) is implied via 'call get_session to poll for status and output.' It stops short of cost, concurrency limits, or failure behavior, 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?

Three short sentences, front-loaded with the core action, then workflow sequencing, then the auth prerequisite. Every sentence earns its place with no filler.

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 7-parameter creation tool with no output schema, the description covers the essential workflow an agent needs: what it does, that results come back via get_session polling, and how to handle authenticated tasks. Return-shape detail is appropriately deferred to get_session, though session limits and expected duration are unaddressed.

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 100%, so all 7 parameters are already documented in the schema, which sets the baseline at 3. The description reinforces profile_id's purpose and cross-references list_browser_profiles, but adds no syntax or format detail beyond what the schema already gives.

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?

States a specific verb and resource ('Create a new browser agent session and run a task') and immediately names where it sits in the workflow relative to get_session and list_browser_profiles. An agent can distinguish it from send_task or stop_session without opening any schema.

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?

Explicit routing: poll with get_session after creation, and use list_browser_profiles first when authentication is required. Both the when-to-use and the prerequisites are stated rather than inferred.

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

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

With no annotations, the description carries the full burden. It usefully discloses the idle-status precondition, but says nothing about what happens if the precondition is violated, whether the call blocks until the task finishes, or how the session state changes afterward (the keep_alive parameter's effect is left entirely to the schema).

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?

Three short sentences, front-loaded with the action and immediately followed by the precondition. The final sentence is slightly redundant with the 'follow-up' framing but still earns its place by naming the reuse-of-context rationale.

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 tool with no annotations and no output schema, the description covers purpose and precondition but omits return/result behavior and error handling. An agent can call it, but cannot predict whether the response is immediate or asynchronous.

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 100%, so all four parameters (task, keep_alive, session_id, output_schema) are already documented in the schema, and the description adds no format or constraint details beyond it. Baseline 3 applies when the schema does the heavy lifting.

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 states a specific verb and resource: 'Send a follow-up task to an existing keep-alive session.' It implicitly separates itself from run_session by emphasizing 'follow-up' on an existing session rather than starting a new one, though it never names the sibling explicitly.

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?

It gives a concrete precondition ('The session must be in idle status') and a clear use case ('multi-turn interactions where you want to reuse the same browser context'). It stops short of naming run_session as the alternative for starting fresh or stating what to do when the session is not idle.

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

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

With no annotations, the description carries the full burden and does disclose the critical behavioral trait: 'session' (the default) destroys the sandbox entirely while 'task' is non-destructive and leaves the session idle. It omits other traits such as whether stopping an already-idle session errors, idempotency, or auth requirements.

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?

Two sentences, front-loaded with the action and then the strategy semantics; no filler and no redundancy. Every clause earns its place by resolving the task-vs-session ambiguity.

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 2-parameter tool with no output schema and no annotations, the description covers the destructive default and the two modes adequately. Return/response behavior and failure modes are absent, but an agent has enough to call it correctly and safely.

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 coverage is 100%, so the baseline is 3, but the description adds meaning beyond the schema text by spelling out the practical consequence of each enum value ('session stays idle for follow-ups') and restating that 'session' is the default. That consequence framing is genuinely useful rather than pure repetition.

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?

States a specific verb and resource ('Stop a running session') that an agent can immediately distinguish from run_session, send_task, and the get_/list_ readers. It stops short of naming a sibling or scoping the tool explicitly, so it is clear but not fully differentiated.

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 explicitly routes between the two strategies: 'task' to stop only the current task and keep the session idle for follow-ups, 'session' (default) to destroy the sandbox. That is concrete when-to-use guidance for the main decision an agent faces, though it does not discuss prerequisites or error conditions.

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