Skip to main content
Glama
escapeWu

chrome-agent-bridge

by escapeWu
OverviewSchemaSourceRelated ServersDiscussionsAdmin
JavaScript
Local

Act on page element

browser_act

Perform one atomic Chrome browser action—click, fill, press, or select—using a current snapshot ref/snapshotId or selector, with confirmation for submitting clicks.

Instructions

Perform one atomic high-level action. Prefer a ref plus snapshotId from the latest browser_snapshot. Clicks run a complete CDP mouse sequence inside one request; press supports common navigation keys; select is for native controls.

Input Schema

TableJSON Schema
NameRequiredDescriptionDefault
keyNoRequired for press, for example ArrowDown, Enter, Escape, Tab, or a single character.
refNoShort-lived ref from the latest browser_snapshot on the same tab.
kindYes
tabIdYes
valueNoRequired for fill; accepted as one value for select.
valuesNoValues or labels for native select.
browserNoWhich connected Chrome to use: an instanceId or label from browser_list_instances. Required when more than one browser is connected; session-based tools infer it from their sessionId.
selectorNoCompatibility fallback when no snapshot ref is available.
confirmedNoSet true only after explicit user confirmation for a potentially submitting click.
snapshotIdNoRequired with ref: the snapshotId returned by the browser_snapshot that produced the ref.
debuggerSessionIdNoRaw or network session ID this task holds on the tab. Required when the tab has a live debugger lease; omit otherwise.

Schema Changelog

Changes observed during successful MCP inspections.

  1. Changed2 schema fields changedv0.10.0
    • addedInput schema / properties / debuggerSessionId
      Added value: +{
+  "description": "Raw or network session ID this task holds on the tab. Required when the tab has a live debugger lease; omit otherwise.",
+  "maxLength": 120,
+  "minLength": 20,
+  "type": "string"
+}
    • addedInput schema / properties / snapshotId
      Added value: +{
+  "description": "Required with ref: the snapshotId returned by the browser_snapshot that produced the ref.",
+  "maxLength": 100,
+  "minLength": 1,
+  "type": "string"
+}
  2. Changed1 schema field changedv0.9.1
    • addedInput schema / properties / browser
      Added value: +{
+  "description": "Which connected Chrome to use: an instanceId or label from browser_list_instances. Required when more than one browser is connected; session-based tools infer it from their sessionId.",
+  "maxLength": 100,
+  "minLength": 1,
+  "type": "string"
+}
  3. Addedv0.8.0

TDQS

B3.1/5.0
Behavior3/5

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 add useful behavior: clicks execute a complete CDP mouse sequence within one request, press handles navigation keys, select targets native controls. But it omits safety-critical behavior the schema hints at — that 'confirmed' gates potentially submitting clicks and that refs are short-lived — and says nothing about error behavior or whether actions wait for navigation.

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?

Three tightly packed sentences, front-loaded with the core action and then the highest-value usage hint (ref + snapshotId) before the per-kind notes. No filler, though the opening abstraction costs a little clarity for the space it occupies.

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

Completeness3/5

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

For an 11-parameter mutating tool with no annotations and no output schema, the description is only partially complete: it never explains the confirmed-gate semantics for submitting clicks, what happens on failure, or what the call returns. The high schema coverage (82%) offsets some of this, but the behavioral and safety context is thin.

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

Parameters3/5

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

Schema description coverage is 82%, so the baseline is 3. The description reinforces the ref+snapshotId pairing and kind-specific intent (press keys, select for native controls), but much of this duplicates schema text such as the key example list, and the fill/value requirement and the confirmed flag's meaning are left to the schema.

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

Purpose3/5

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

The description opens with a vague abstraction ('Perform one atomic high-level action') and never states plainly that this clicks/fills/presses/selects a page element, though the follow-up clauses about CDP mouse sequences, navigation keys, and native <select> controls do clarify the domain. It also fails to distinguish itself from the sibling tools browser_click and browser_fill, which appear to overlap directly, leaving an agent unsure which entry point to choose.

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

Usage Guidelines3/5

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

The 'Prefer a ref plus snapshotId from the latest browser_snapshot' line is genuine workflow guidance implying browser_snapshot must run first, and the schema adds a selector fallback. However, there is no tool-level when-to-use/when-not guidance, and with browser_click and browser_fill as siblings the omission of any routing rule is a real gap.

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

Deploy Server

Other Tools