Skip to main content
Glama

cortex_run

Destructive

Executes batched Chromium browser actions and assertions, supporting targeted reads, CSS/ref selection, expansion, extraction, and recovery from stale or ambiguous targets.

Instructions

Azioni/assert e piani each/expand/extract. Target: ref:s1-e1 o s1-e1; CSS; role:button|Nome, label:Nome. each/extract solo CSS; each limit max 50; fields maxLength 1..500. 25 step/200 azioni/60s. StaleRef/AmbiguousTarget: step non eseguito, max 5 candidati correnti; scegli il controllo e riprendi dallo step fallito senza ripetere quelli completati.

Tool call template

Replace each placeholder with a value matching the input schema. Only required parameters are included; add optional parameters as needed.

{
  "id": 1,
  "jsonrpc": "2.0",
  "method": "tools/call",
  "params": {
    "arguments": {
      "steps": "<steps>"
    },
    "name": "cortex_run"
  }
}

Input Schema

NameRequiredDescriptionDefault
frameNo
stepsYes
sessionIdNoID restituito da newSession. Omettere per usare la sessione condivisa.
operationIdNo
observeAfterNo
project_pathNo
allowDestructiveNo

Schema Changelog

Changes observed during successful MCP inspections.

  1. First observedv0.2.0-dev.1

TDQS

B3.4/5.0
Behavior4/5

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

Annotations already flag destructive=true, openWorld=true and non-idempotent, so the bar is lower, yet the description adds substantial behavior: hard rate/volume limits, and the exact failure semantics for StaleRef/AmbiguousTarget (step not executed, at most 5 current candidates, resume from the failed step without re-running completed ones). It does not mention that destructive actions need allowDestructive, nor what the response contains, which keeps it from a 5.

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

Conciseness3/5

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

It is short with no filler, but it is telegraphic and semicolon-chained, opening with operation names rather than what the tool is or when to call it. The recovery behavior, arguably the most important sentence, is buried last. Information density is high but parseability is low.

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 a complex, destructive, open-world tool with 25 nested steps and no output schema, the description covers the hardest parts (target syntax, caps, stale-ref recovery) but omits the destructive gate, session handling, and any hint of what is returned. An agent could invoke it but would be guessing on several top-level parameters.

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?

With only 14% schema coverage, the description has to carry parameter meaning, and it does: target grammar (ref:s1-e1 or s1-e1, CSS, role:|label:, each/extract CSS-only), each limit and fields maxLength. One discrepancy: it says each limit max 50 while the schema permits up to 100, which could mislead an agent. It also says nothing about sessionId, operationId, observeAfter, allowDestructive or project_path.

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 names the operation families it runs ("Azioni/assert e piani each/expand/extract") and the target syntaxes it accepts, so an agent can infer this executes a browser step plan. However, it never states plainly that it runs the `steps` array against a page, and it uses dense internal jargon rather than a clear verb+resource phrasing. It is distinguishable from cortex_peek/cortex_open only by implication.

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?

It gives real constraints on how to use it: each/extract require CSS-only targets, each is capped at 50, fields at 1..500 chars, and 25 steps/200 actions/60s. What is missing is explicit when-to-use guidance versus siblings (e.g. use peek first to obtain refs) and when not to call it; the recovery instruction for StaleRef implies a peek-then-resume loop but never names it.

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