Skip to main content
Glama

browser_upload_file

Assigns absolute browser-host file paths to a file input or pending chooser, supports clearing/canceling selections, and verifies the retained FileList count; multiple files require a multiple input.

Instructions

Set browser-host absolute file paths on one file-input/label ref/selector, or omit target for the pending chooser. [] clears/cancels; multiple files need a multiple input; directories unsupported. assignment_acknowledged confirms the command, not event delivery/upload. count_verified compares retained FileList count (null if unreadable); mismatch is not assignment failure. Observe before uploading again.

Input Schema

NameRequiredDescriptionDefault
nthNoSelector only: zero-based DOM index, AX tree order for role.
refNo
nameNorole only: exact AX name; empty allowed.
filesYes
frameNoChild frame: at least one of url/title/path. url/title exact; path=unique nested CSS iframe steps (e.g. ["#frame"]); exclusive with frame_id.
observeNoRead this explicit control after the action within its remaining budget; observation is separate from delivery. No assertion or automatic replay.
frame_idNoSelector frame; default main.
selectorNoQuery by selector_type; role example: selector="link", name="Claim".
tab_handleYes
timeout_msNoTotal preparation/dispatch/verification and optional observation budget.
dialog_policyNoaccept/dismiss auto-handle dialogs and return their messages.ask
selector_typeNoDefault css. label: exact associated HTML label, whitespace normalized. placeholder: exact attribute.

Schema Changelog

Changes observed during successful MCP inspections.

  1. Changed2 schema fields changedv0.28.2
    • addedInput schema / properties / observe
      Added value: +{
      +  "additionalProperties": false,
      +  "description": "Read this explicit control after the action within its remaining budget; observation is separate from delivery. No assertion or automatic replay.",
      +  "properties": {
      +    "frame": {
      +      "additionalProperties": false,
      +      "description": "Child frame: at least one of url/title/path. url/title exact; path=unique nested CSS iframe steps (e.g. [\"#frame\"]); exclusive with frame_id.",
      +      "properties": {
      +        "path": {
      +          "items": {
      +            "minLength": 1,
      +            "type": "string"
      +          },
      +          "maxItems": 16,
      +          "minItems": 1,
      +          "type": "array"
      +        },
      +        "title": {
      +          "minLength": 1,
      +          "type": "string"
      +        },
      +        "url": {
      +          "minLength": 1,
      +          "type": "string"
      +        }
      +      },
      +      "type": "object"
      +    },
      +    "frame_id": {
      +      "description": "Selector frame; default main.",
      +      "type": "string"
      +    },
      +    "max_chars": {
      +      "default": 16000,
      +      "maximum": 64000,
      +      "minimum": 1000,
      +      "type": "integer"
      +    },
      +    "max_options": {
      +      "default": 30,
      +      "maximum": 100,
      +      "minimum": 1,
      +      "type": "integer"
      +    },
      +    "name": {
      +      "description": "role only: exact AX name; empty allowed.",
      +      "maxLength": 10000,
      +      "type": "string"
      +    },
      +    "nth": {
      +      "description": "Selector only: zero-based DOM index, AX tree order for role.",
      +      "maximum": 9007199254740991,
      +      "minimum": 0,
      +      "type": "integer"
      +    },
      +    "popup_selector": {
      +      "description": "Optional unique CSS popup in the control frame; otherwise use browser-resolved controls/owns relations and control descendants.",
      +      "maxLength": 10000,
      +      "minLength": 1,
      +      "type": "string"
      +    },
      +    "ref": {
      +      "type": "string"
      +    },
      +    "selector": {
      +      "description": "Query by selector_type; role example: selector=\"link\", name=\"Claim\".",
      +      "maxLength": 10000,
      +      "minLength": 1,
      +      "type": "string"
      +    },
      +    "selector_type": {
      +      "description": "Default css. label: exact associated HTML label, whitespace normalized. placeholder: exact attribute.",
      +      "enum": [
      +        "css",
      +        "xpath",
      +        "label",
      +        "placeholder",
      +        "role"
      +      ],
      +      "type": "string"
      +    },
      +    "timeout_ms": {
      +      "default": 1000,
      +      "maximum": 5000,
      +      "minimum": 100,
      +      "type": "integer"
      +    }
      +  },
      +  "type": "object"
      +}
    • changedInput schema / properties / timeout_ms / description
      Previous value: -"Total preparation/dispatch/verification budget."New value: +"Total preparation/dispatch/verification and optional observation budget."
  2. Changed8 schema fields changedv0.24.2
    • changedInput schema / properties / dialog_policy / description
      Previous value: -"accept/dismiss returns handled dialog messages."New value: +"accept/dismiss auto-handle dialogs and return their messages."
    • changedInput schema / properties / frame / description
      Previous value: -"Exact child-frame URL/title or CSS iframe path (e.g. [\"#content_iframe\"]); alternative to frame_id. Each path entry uniquely selects an iframe in the preceding document."New value: +"Child frame: at least one of url/title/path. url/title exact; path=unique nested CSS iframe steps (e.g. [\"#frame\"]); exclusive with frame_id."
    • changedInput schema / properties / frame_id / description
      Previous value: -"Selector scope; defaults to main frame."New value: +"Selector frame; default main."
    • changedInput schema / properties / name / description
      Previous value: -"role only: exact accessible name; may be empty."New value: +"role only: exact AX name; empty allowed."
    • changedInput schema / properties / nth / description
      Previous value: -"Zero-based match index: DOM order, or AX tree order for role."New value: +"Selector only: zero-based DOM index, AX tree order for role."
    • changedInput schema / properties / selector / description
      Previous value: -"CSS/XPath query, HTML label text, exact placeholder or AX role (e.g. selector=\"link\", name=\"Claim\"), according to selector_type."New value: +"Query by selector_type; role example: selector=\"link\", name=\"Claim\"."
    • changedInput schema / properties / selector_type / description
      Previous value: -"Default css. label matches associated HTML labels with whitespace normalized; role uses browser AX semantics."New value: +"Default css. label: exact associated HTML label, whitespace normalized. placeholder: exact attribute."
    • changedInput schema / properties / timeout_ms / description
      Previous value: -"Covers preparation, dispatch and verification."New value: +"Total preparation/dispatch/verification budget."
  3. Changed1 schema field changedv0.24.1
    • changedInput schema / properties / selector / description
      Previous value: -"CSS/XPath query, HTML label text, exact placeholder or AX role, according to selector_type."New value: +"CSS/XPath query, HTML label text, exact placeholder or AX role (e.g. selector=\"link\", name=\"Claim\"), according to selector_type."
  4. First observedv0.24.0

TDQS

A4/5.0
Behavior4/5

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

With no annotations, the description carries the full burden and does substantial work: it distinguishes assignment_acknowledged from event delivery and from actual upload, and explains that count_verified compares the retained FileList and that a mismatch is not an assignment failure. Still missing: whether paths must exist/be readable by the host, and what happens on timeout or failure.

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

Conciseness4/5

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

Front-loads the action, then the target-omission case, then constraints and result-interpretation. Every clause carries information, but the ack/count_verified sentences are compressed into telegraphic fragments that take a second read.

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

Completeness4/5

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

For a 12-parameter, nested-object mutation tool with no annotations and no output schema, the description covers the essential semantics: targeting, clearing, multi-file eligibility, and how to interpret the acknowledgement and verification fields. Remaining gaps (error/timeout behavior, path validity preconditions) are secondary.

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

Parameters4/5

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

Schema coverage is 75%, and the description adds semantics the schema does not: files must be host-absolute paths, [] means clear/cancel, >1 file needs a multiple input, directories are rejected, and the target trio (ref/selector/omit) maps to specific behaviors. It says nothing about nth, name, frame/frame_id, timeout_ms or dialog_policy, leaving those to the schema.

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

Purpose4/5

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

Opens with a specific verb and resource: set browser-host absolute file paths on one file-input/label ref/selector, or on the pending chooser. That is unambiguous and clearly distinct from browser_click/browser_fill/browser_type siblings, though it never names those siblings to sharpen the boundary.

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

Usage Guidelines4/5

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

Gives concrete situational rules: omit target for the pending chooser, pass [] to clear/cancel, multiple files require a multiple input, directories unsupported, and 'observe before uploading again'. It stops short of stating the upstream trigger (e.g. that the chooser must first be opened by a click) or naming an alternative tool.

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