Skip to main content
Glama

Observe Native Calls

observe_native_calls
Destructive

Launch the active Mach-O under LLDB to auto-record entries into chosen functions or Objective-C methods, capturing thread, module, addresses, arguments, selector, receiver class, and caller frames.

Instructions

Launch the active Mach-O as an owned process under LLDB and record each entry into caller-selected functions or Objective-C methods: thread, module, symbol, load and file address, raw argument registers, selector, receiver class and optional caller frames. Every hit auto-continues. No expression is evaluated and no other process is attached. The process is killed when max_events or duration_ms is reached; exit status, captured stdout/stderr, unresolved breakpoints and limitations are returned inline. The process runs with the current user's permissions and may change files, show UI or use the network. A hardened-runtime target needs the get-task-allow entitlement.

Example tool call

This call uses example arguments supplied in the tool’s input schema.

{
  "id": 1,
  "jsonrpc": "2.0",
  "method": "tools/call",
  "params": {
    "arguments": {
      "arguments": [
        "--help"
      ],
      "breakpoints": [
        {
          "kind": "objc-method",
          "selector": "dataTaskWithRequest:completionHandler:",
          "class_name": "NSURLSession"
        },
        {
          "kind": "function",
          "name": "open",
          "module": "libsystem_kernel.dylib"
        }
      ],
      "duration_ms": 5000,
      "backtrace_frames": 4
    },
    "name": "observe_native_calls"
  }
}

Input Schema

NameRequiredDescriptionDefault
argumentsNo
max_eventsNoStop observing and kill the process after this many recorded entries.
breakpointsYes
duration_msNoObservation window; the process is killed if it is still running.
environmentNoOverrides on top of the environment REA runs with.
backtrace_framesNoCaller frames to record per entry.
working_directoryNo
argument_registersNoInteger argument registers to record per entry (arm64 has 8, x86_64 has 6).

Schema Changelog

Changes observed during successful MCP inspections.

  1. First observedv0.1.0

TDQS

A4.4/5.0
Behavior5/5

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

Annotations already flag destructive/openWorld/non-idempotent, and the description goes well beyond them: auto-continue on every hit, process killed at max_events or duration_ms, exit status plus captured stdout/stderr and unresolved breakpoints returned inline, runs with current user's permissions and may mutate files/show UI/use the network, and hardened-runtime targets need the get-task-allow entitlement. This is exactly the risk and lifecycle context an agent needs before launching a destructive debug session.

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

Conciseness5/5

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

Four sentences, highly front-loaded: what it launches first, then capture fields, then lifecycle/termination, then safety and entitlement caveats. Heavy noun lists are dense but every clause carries signal; nothing is padding.

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?

With no output schema and 8 parameters, the description correctly spends effort on return contents (entries, exit status, stdout/stderr, unresolved breakpoints, limitations) and on the destructive process lifecycle. The remaining gap is minor but real: working_directory and environment overrides that meaningfully change execution are not addressed anywhere.

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 coverage is 63%, and the description's parameter content largely restates existing schema text (kill on max_events/duration_ms, argument registers, caller frames). It says nothing about the required breakpoints selection shape, working_directory (which has no schema description at all), environment overrides, or arguments — so it does not compensate for the uncovered parameters.

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

Purpose5/5

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

States a specific verb chain and resource: launch the active Mach-O under LLDB as an owned process and record entries into caller-selected functions/Objective-C methods. The enumerated capture fields (thread, module, symbol, addresses, registers, selector, receiver class, frames) make the output shape unmistakable and distinguish it from inspect_*/trace_* siblings.

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?

The description clearly frames the context of use (dynamic observation of call entry under a debugger) and implicitly separates itself from expression-evaluating tools by stating 'No expression is evaluated and no other process is attached.' However, it never names a sibling alternative or states when NOT to use it (e.g. when static analysis via inspect_native_api would suffice).

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