Skip to main content

Documentation Index

Fetch the complete documentation index at: https://mintlify.com/wtyler2505/ProtoPulse/llms.txt

Use this file to discover all available pages before exploring further.

ProtoPulse’s AI assistant translates your natural language directly into tool actions. When you type a message, the AI parses your intent, selects the appropriate tools from the 82-action registry, constructs validated parameters, and executes the actions against your design. Understanding how that mapping works helps you write prompts that get precise, consistent results — especially for complex multi-step operations.

​
How the AI interprets your text

The AI does not just answer questions — it decides which tool actions to call and in what order. A single message can trigger a chain of five or ten tool calls. The AI receives your message alongside the full current state of your project: your architecture nodes and edges, BOM, validation issues, schematic sheets, and design history. It uses that context to make sensible decisions about what to place, how to connect things, and what to name nets. When the AI is uncertain about a parameter (for example, you asked it to “add a capacitor” but did not specify a value), it either picks a sensible default and tells you, or asks a clarifying question before acting. Providing specifics upfront produces more accurate first-pass results.

​
Prompt patterns

The most powerful use of the AI is generating an entire architecture from a single description. The generate_architecture tool can create 10+ nodes and all their connections in one call.Pattern: Describe the system with key specs, connectivity, and any constraints.
good prompt
Design an IoT weather station with an ESP32 as the main MCU,
a BME280 for temperature/humidity/pressure over I2C, a LoRa
SX1276 module for 915MHz transmission over SPI, and a 3.7V
LiPo battery with a TP4056 charging circuit. Include a 3.3V
LDO for power regulation.
The AI will place all nodes with appropriate types, draw typed signal edges (I2C, SPI, Power), and label each connection with the correct signal name.What to include:
  • The MCU or main processor
  • Sensors and peripherals with their bus types (I2C, SPI, UART)
  • Power supply topology (input voltage, output rail, regulator type)
  • Any special interfaces (USB, CAN, Ethernet)
After generating the architecture, follow up with “populate the BOM with Digi-Key pricing” to add real part numbers and costs in one additional step.
When you want to add one block to an existing design without regenerating everything, be specific about where it connects and what it does.Pattern: Name the component, its purpose, and how it connects to existing blocks.
good prompt
Add an SD card module connected to the ESP32 via SPI. Use the
same SPI bus as the LoRa module. The chip select should be a
separate GPIO pin.
less specific
Add an SD card.
The specific version tells the AI which existing node to connect to, which bus to use, and the pin topology. The vague version will work, but the AI has to guess the connection details and may need a follow-up correction.
Ask the AI to run a specific type of check rather than a generic “check my design.” Targeted prompts produce more actionable findings.Pattern: Name the specific concern you want checked.
good prompts
Run a power budget analysis for this design.

Check for voltage level mismatches between the ESP32 (3.3V logic)
and any 5V peripherals.

Run DFM checks against JLCPCB design rules.

What ESD protection is missing for the USB-C port?
less useful
Check my design.
The AI can call power_budget_analysis, voltage_domain_check, dfm_check, and run_drc independently. Specifying which analysis you want targets the right tool and gives you a focused response.
After validation, follow up with “auto-fix the warnings” to trigger the auto_fix tool, which adds missing decoupling capacitors, pull-up resistors, and ESD protection components automatically.
Give the AI pricing and availability constraints upfront so it can make sourcing decisions in the first pass rather than requiring corrections.Pattern: Specify distributor, budget constraints, or availability requirements.
good prompt
Add all components to the BOM with Digi-Key pricing. Prefer
parts that are in stock, and keep the total BOM cost under $15
for a 10-unit run.
alternative
Export the BOM as a JLCPCB-compatible CSV. Flag any parts that
are not available in JLCPCB's basic parts library.
For parametric searches, be specific about electrical specifications:
good prompt
Find a 100nF ceramic capacitor in 0402 package, rated for at
least 10V, preferably X7R dielectric, available from LCSC.
Export prompts are straightforward — name the format and any relevant options.
export prompts
Export this design to KiCad.

Generate Gerber files for JLCPCB — I need copper layers,
solder mask, silkscreen, and drill files.

Export a SPICE netlist for LTspice.

Generate a pick-and-place CSV for automated assembly.

Create a full design report as a Markdown file.
The AI calls the appropriate export tool, generates the file server-side, and triggers a browser download. No extra steps are required.
You can request multiple exports in one message: “Export the KiCad schematic, a SPICE netlist, and a BOM CSV.” The AI will call each export tool in sequence and download all three files.
The AI can explain design choices, component selections, and net connections. These prompts do not trigger tool actions — they use the AI’s knowledge and your project context to answer questions.Pattern: Reference specific parts of your design.
explanation prompts
Why is there a 100nF capacitor between VCC and GND near the ESP32?

Explain the purpose of the ferrite bead between the USB input
and the 5V power rail.

What is the SPI_CS net connected to, and why?

Should I use a common-mode choke or a ferrite bead for this
USB differential pair?
The AI has access to your full design context and will reference your specific nodes, edges, and BOM items in its explanation.

​
Example prompt showcase

The following prompts represent complete, realistic design tasks. Each demonstrates a different area of ProtoPulse’s AI capabilities. 
Design a 3.3V regulated power supply from USB-C input with input
protection and decoupling. Include a TVS diode for ESD protection,
a ferrite bead for EMI filtering, and an LDO regulator with input
and output decoupling capacitors. Then populate the BOM with
Digi-Key part numbers and run a power budget analysis.

​
When the AI makes a mistake

The AI will occasionally place the wrong component, use the wrong bus type, or generate a BOM entry with an incorrect part number. Here is how to recover efficiently. Undo immediately: If you spot a mistake right away, type “undo that” or press Ctrl+Z. The AI calls the undo tool and reverses the last action. Be specific in your correction: Instead of “that’s wrong, redo it,” describe exactly what is incorrect and what you want instead.
corrective prompt
The SPI bus between the MCU and LoRa module should use GPIO 18 for
SCLK, GPIO 19 for MISO, and GPIO 23 for MOSI — not the I2C pins.
Update the edge label to reflect the correct pin assignments.
Use the Audit Trail: The Audit Trail view (switch_view → audit_trail) shows every AI action with its exact parameters. If you are unsure what changed, review the log before deciding how to correct it. Break complex corrections into steps: If a generated architecture has multiple issues, correct them one at a time rather than writing a long compound instruction. The AI is more reliable when each message targets one change.

​
Attaching images

You can attach an image to any message using the camera/image upload button in the chat panel. The AI uses its vision capabilities (via the analyze_image tool) to interpret the image. Useful image types to attach:
  • Circuit photos — the AI identifies components on a real PCB and can suggest equivalent schematic representations
  • Hand-drawn schematics — the AI extracts component types and connections from a whiteboard sketch
  • Reference datasheets — attach a pinout diagram and ask the AI to set pin mappings for a node
  • Existing PCB layouts — ask the AI to identify potential design issues visible in the photo
image prompt examples
[attach photo of PCB]
Identify the main components on this board and add them to the BOM.

[attach hand-drawn schematic]
Recreate this schematic in ProtoPulse as an architecture diagram.

[attach ESP32 pinout image]
Set the pin map for the ESP32 node using this pinout.

​
Iterative design tips

Start broad, then refine. Generate a complete architecture first, then use follow-up prompts to adjust individual nodes, correct bus types, add missing components, and tune the BOM. Trying to specify every detail in the first prompt often produces a longer response that is harder to review.
Use the custom system prompt (AI Settings → Custom System Prompt) for standing constraints that apply to every message. For example: “Always prefer STM32 over ESP32 for industrial designs” or “Use only components available on LCSC with stock above 1000 units.” This saves you from repeating constraints in every prompt.
Reference your project’s existing nodes by their exact label. If you have a node labeled “ESP32-S3,” use that exact name in your prompt — “connect a BME280 to the ESP32-S3 via I2C.” The AI uses node labels to look up the correct connection targets.
Ask for explanations after generating a design. “Explain the power topology of this design” or “Why did you choose the TP4056 over the MCP73831?” gives you insight into the AI’s reasoning and helps you catch any incorrect assumptions before they propagate into the BOM and schematic.

Build docs developers (and LLMs) love

Get started for freeTalk to us