Prompts

mcgpib provides 5 MCP prompts that encode domain knowledge about instrument initialization, troubleshooting, and measurement workflows. Prompts return structured step-by-step instructions that guide the LLM through multi-tool sequences, including fallback strategies for common failure modes.

---

initialize_bench

Connect to a bridge, scan the bus, and identify all instruments. Use this when setting up a test bench for the first time or after power cycling instruments.

Parameters

Name	Type	Required	Description
bridge_name	string	Yes	Name of the bridge to initialize

Workflow

1. Call connect_bridge to establish communication
2. Call bus_scan to discover instruments
3. Note each instrument’s address and identity
4. Summarize the bench setup (firmware version, instrument count, unidentified addresses)

Includes troubleshooting guidance for:

• Serial transport connection failures (USB cable, port permissions)
• TCP transport connection failures (WiFi, IP address, single-client limit)
• Empty bus scans (cable issues, instrument power, bus reset via interface_clear)

When to use

At the start of any session, or after instruments have been added, removed, or power cycled.

Example

prompt = await client.get_prompt("initialize_bench", {
    "bridge_name": "bench-a"
})

---

troubleshoot_instrument

Systematic 6-step diagnostic sequence for a non-responsive instrument. Walks through bus connectivity, identification, status polling, error clearing, reset, and bus-level recovery.

Parameters

Name	Type	Required	Description
bridge_name	string	Yes	Name of the bridge
address	integer	Yes	GPIB address of the problematic instrument

Workflow

1. Verify bus connectivity — bus_scan with identify=false to check if the address appears as a listener
2. Check identity — instrument_identify to test *IDN? support
3. Check status — serial_poll to read the status byte (RQS, ESB, MAV bits)
4. Clear error state — Send *CLS, then loop SYST:ERR? until 0,No error
5. Reset if needed — instrument_reset followed by re-identification
6. Bus-level recovery — interface_clear and bus_clear as last resort

When to use

When an instrument stops responding to commands, returns unexpected data, or was working previously but has become unreliable.

Example

prompt = await client.get_prompt("troubleshoot_instrument", {
    "bridge_name": "bench-a",
    "address": 22
})

---

measurement_sequence

Guide setup and capture of a specific measurement type. Covers instrument identification, configuration, triggering, reading, and validation.

Parameters

Name	Type	Required	Description
bridge_name	string	Yes	Name of the bridge
address	integer	Yes	GPIB address of the measurement instrument
measurement_type	string	Yes	Type of measurement (e.g., dc_voltage, frequency, resistance, ac_voltage, current_dc, 4wire_resistance)

Workflow

1. Identify the instrument — determine model and available SCPI commands
2. Configure the measurement — send CONF:* commands appropriate to the measurement type and instrument model
3. Trigger and read — either READ? for single-shot or the INIT / FETCH? pattern for triggered measurements
4. Validate — check the returned value and instrument status byte

Includes a table of common SCPI configuration commands mapped to measurement types:

Type	SCPI Pattern
DC Voltage	CONF:VOLT:DC [range],[resolution]
AC Voltage	CONF:VOLT:AC [range],[resolution]
Frequency	CONF:FREQ [range],[resolution]
Resistance	CONF:RES [range],[resolution]
4-Wire Resistance	CONF:FRES [range],[resolution]
DC Current	CONF:CURR:DC [range],[resolution]

When to use

When you need to take a specific measurement and want guidance on the correct command sequence for the instrument at hand.

Example

prompt = await client.get_prompt("measurement_sequence", {
    "bridge_name": "bench-a",
    "address": 22,
    "measurement_type": "dc_voltage"
})

---

bus_health_check

Full infrastructure health audit across all configured bridges. Checks every bridge’s connectivity, scans every bus, polls every instrument, and generates a consolidated health report.

Parameters

This prompt takes no parameters.

Workflow

1. List all bridges — note connected vs. disconnected
2. For each connected bridge — run bus_scan with identification, check_srq, and serial_poll for all instruments
3. For each disconnected bridge — attempt connect_bridge and log results
4. Generate report — total bridges (configured, connected, failed), total instruments (discovered, identified, responding), SRQ assertions, error status bytes, and recommendations

When to use

Periodic health checks, morning bench setup verification, or after a power event that may have affected multiple instruments.

Example

prompt = await client.get_prompt("bus_health_check", {})

---

scpi_command_help

Instrument-specific SCPI command reference. Given a model identifier, provides the common measurement commands, status/error handling patterns, and any non-standard behavior notes.

Parameters

Name	Type	Required	Description
instrument_model	string	Yes	Model identifier (e.g., HP3478A, Keithley 2000, Agilent 34401A)

Workflow

1. Measurement commands — CONF:*, SENS:*, TRIG:*, READ?, FETCH?, MEAS:*, range/resolution settings
2. Status and error handling — status byte bit assignments, SYST:ERR?, event status register
3. Important notes — IEEE 488.2 compliance, non-standard GPIB behavior, typical timeout requirements, post-power-on initialization

When to use

When working with an unfamiliar instrument model, or when you need to go beyond basic MEAS:* commands.

Note

Older instruments (HP 3478A, Fluke 8842A, Keithley 195A) often use proprietary command sets rather than standard SCPI. The prompt accounts for this and will note when an instrument predates the SCPI standard.

Example

prompt = await client.get_prompt("scpi_command_help", {
    "instrument_model": "HP 34401A"
})