If you used an AI assistant to build a Fulcrum solution and something isn't working as expected, you can ask the same assistant to diagnose and fix it. A clear, specific prompt helps the AI find the cause faster and make a smaller, safer change. This article gives you copy-ready prompts for common problems with apps, reports, app extensions, Data Events, calculations, the Fulcrum MCP connection, and integrations.
You don't need to know internal field names or APIs. Ask the AI to discover them from your live configuration.
If your AI assistant can't connect to Fulcrum, see Troubleshooting Fulcrum MCP and the AI Toolkit.
Using the prompts in this article
Return to the AI conversation where you built your solution.
Find your issue in this article, then copy the matching prompt.
Replace anything in brackets, such as
[task], with your details.Describe what you expected, what happened instead, and the steps that caused the problem. Include screenshots or the exact error message when available.
If you start a new conversation, include the latest implementation and a short summary of what you already found.
Redact credentials, such as API tokens, and any unnecessary personal information before you share details with your AI assistant. Never paste your API token into a chat.
Giving the AI enough context
The more context you provide, the more accurate the fix. For each problem, include:
Target: The app or report name or link, and whether you're using a test or production environment.
Conditions: The browser or device, whether you're online or offline, and the version.
Evidence: One failing example and, if possible, one working example.
Scope: What the AI may change.
Success: The result that will prove the fix.
Keywords that help the AI find the right tools
Use these terms naturally with your goal. They are routing cues, not guaranteed commands. The tools and permissions available to your AI assistant still determine what it can do.
Your task | Helpful keywords |
Fix an app or rule | App design; Record Links; repeatables; Data Events; trigger; calculations |
Fix a custom screen | App Extension; Reference File; extension bridge; save/cancel; field mapping |
Fix a report or summary | Advanced Report Template; PDF or HTML; pagination; Query API; related records |
Connect another system | REST API; native Workflow; webhook; authentication; retries; duplicate prevention |
Diagnose and verify | Reproduce; exact error; last working version; smallest fix; end-to-end test; save/reopen; live read-back |
Troubleshooting apps
Start with what one record represents, who uses it, and the task they need to complete.
Useful keywords: live form schema, app design, repeatables, Record Links, stored values
Question or issue | Prompt to copy |
The app does not fit my workflow | Walk through this app as a field user completing [task]. Compare it with my original requirements. Identify what needs to change before making edits. Clarify what one record represents and which activities repeat. |
Why is this field hidden, required, or read-only? | Check this field's settings and any rules or code affecting it. Explain what causes this behavior before changing it. |
Why are dropdown choices missing or incorrect? | Check the configured choices and any filtering rules. I should see [expected options], but I see [actual options]. Distinguish displayed labels from stored choice values. |
Why are repeating entries not working correctly? | Review this repeating section. Help me test adding, editing, saving, and reopening multiple entries. Check for duplicate rows after applying the same change twice, and preserve existing row identities and unrelated values. |
Why are connections between apps not working? | Explain which app holds the Record Link, what it connects to, and how users select or create related records. Check the settings and required access. |
Will this change affect existing data? | Inspect dependencies in records, calculations, Data Events, reports, and integrations. Preserve existing field keys and data names. Explain any replacement, deletion, or migration before making that change. |
Why do I still see the old version? | Read back the saved app configuration. Check whether I need to refresh the full records page, close and reopen the record, or sync mobile to load it. Verify the version before making another code change. |
How do I verify the fix? | Give me a short walkthrough to test creating, saving, and reopening a record in an approved test context. Preserve existing data and features that already work. Identify any test that would create or change a live record. |
Add this to any change request: "Inspect the current configuration first. Make the smallest justified change, preserve unrelated behavior, and validate the saved result."
Troubleshooting custom reports
Share a sample output and identify the record and template used to generate it. Specify PDF or HTML, paper size, orientation, and how multiple photos should be arranged.
Useful keywords: advanced Report Template, Report Builder, PDF/HTML, pagination, Query API, visual inspection
Question or issue | Prompt to copy |
Why does the report fail or come out blank? | Check the saved template against the current Fulcrum report documentation. Find the smallest section causing the failure. Inspect the returned content type rather than relying only on the download filename. |
Why is information missing or incorrect? | Compare this report with the saved record. Check field references, stored values, blank values, and conditions controlling what appears. Treat missing information as Not recorded unless the business rule says otherwise. |
The extension looks right, but the report is wrong | Trace one example from the extension through its returned result, the native field, the saved record, and the report. Identify the first mismatch; do not assume the report template is the only cause. |
Why are repeating entries missing or duplicated? | Check how the report reads and displays repeating entries. Explain why these rows are missing or duplicated, including any joins or grouping used to retrieve them. |
Why do photos or signatures not appear? | Check the media references and documented report URL helpers. Determine whether the problem is the source data, access, or the template; a media identifier is not itself an image URL. |
Why are text, tables, or images cut off? | Review page size, spacing, table widths, image sizes, and page breaks. Keep headers aligned with the content and repeat them where needed. Inspect rendered pages, not just extracted text. |
Why are totals or included records wrong? | Explain which records and values contribute to this result, including filters, date boundaries, related records, and grouping. Compare them with this expected example: [example]. |
How do I verify the fix? | Generate and review reports with short and long content, multiple repeating entries/photos, and missing optional information. Verify a live Fulcrum-generated result. State which template is installed online and which file is only a preview. |
To learn more about report functions, see the Report Builder functions reference in the Fulcrum Developer Docs.
Troubleshooting app extensions
Use these prompts for custom screens, pickers, and other interactive features opened inside a record. State which devices must work and whether the feature must work offline.
Useful keywords: App Extension, Reference File, OPENEXTENSION, extension bridge, field mapping, save/reopen
Question or issue | Prompt to copy |
Why does the button not open the extension? | Check the button action, opening code, and referenced file. Verify the exact Reference File name and the current extension bridge. Identify where the process stops, including any unintended extra navigation. |
Why does it open as a blank screen? | Check page initialization, incoming data, and required files. Identify anything missing or failing to load. Compare with the current supported extension pattern. |
Why does existing record information not appear? | Compare the information sent from the record with what the extension expects. Check mismatched names, stored value shapes, and data types, especially choices and repeatables. |
Why does my result not return to the record? | Trace the returned result through the receiving Data Event into the target field. Then save and reopen the record. Distinguish a changed graphic, an updated open form, and persisted record data. |
Why does it work online but fail offline? | Identify every internet dependency, including scripts, fonts, images, and API calls. A locally available page does not prove remote data is available offline. Explain what must change for [intended offline use]. |
Why does it work on one device but not another? | Prepare a comparison test for [working device] and [failing device]. Include screen size, touch behavior, scrolling, zoom, and any sandbox restrictions relevant to the feature. |
Why are edits lost or duplicated? | Test Apply, Cancel, close, reopen, and repeated Apply. Preserve existing rows and native edits. Explain which changes are temporary and which require saving the Fulcrum record. |
Why is access denied when the page opens? | Verify authentication and access to one actual resource in this execution environment. Do not assume an extension or external page inherits the main Fulcrum browser session. Keep credentials out of client code and shared URLs. |
How do I verify the fix? | Test opening, loading values, completing an action, canceling, saving, and reopening. Include required devices and offline conditions. Distinguish local bridge tests from live persistence and physical-device tests. |
To learn more, see the App Extensions introduction in the Fulcrum Developer Docs.
Troubleshooting Data Events
Use these prompts for automatic actions such as setting values, changing choices, showing fields, or checking information before saving. Specify when the action should happen and which exceptions take priority.
Useful keywords: Data Events, trigger, field change, new/edit/save, current script, built-in functions
Question or issue | Prompt to copy |
Why does the automatic action not happen? | Check what triggers this action and which fields it references. Explain when it should run and why it is not running in this example: [example]. Verify that the data it needs is available at that moment. |
Why does it work only after I change a field? | Check whether the rule also runs when new and existing records open, including records with default values. |
Why does one value update but the next action fail? | Trace the sequence. Check whether the code incorrectly assumes that setting a value automatically triggers another action, or that a lookup finishes immediately. |
Why is saving blocked unexpectedly? | Identify the rule preventing save. Explain which condition fails using this example: [example]. Distinguish required missing information from an incorrect validation condition. |
Why does a lookup or request fail? | Check inputs, response handling, connectivity, and access requirements. Show the actual error. Check browser-specific request restrictions and whether the chosen event supports the required asynchronous work. |
Why did new code break an existing feature? | Read the complete current saved script. Check for conflicting, overwritten, or duplicated handlers. Preserve unrelated behavior and avoid replacing newer edits with an old local copy. |
Why does automation override my decision? | Define which rule takes priority. For example, auto-complete only Needs Inspection; preserve manually selected Needs Revisit or Could Not Complete. Specify what happens if a required answer is later cleared. |
How do I verify the fix? | Test creating, editing, saving, and reopening records as relevant, including exceptions and blank values. Check whether a built-in setting or calculation meets the requirement more simply. Read back the saved script after updating it. |
Make the business rule concrete. For example: "When Pipe Blockage is Yes, require findings. A blank answer is not No. Do not clear an existing finding when the selection changes unless I request it."
To learn more, see the Data Events reference in the Fulcrum Developer Docs, or What are data events?
Troubleshooting calculations
Provide sample inputs and the result you expect. Include units, precision, and the intended treatment of missing information.
Useful keywords: calculation, field data names, stored choice values, units, rounding, expected results
Question or issue | Prompt to copy |
Why is the result incorrect? | Show the calculation step by step using [inputs]. I expect [result]. Check the formula and field references. |
Why is the result blank or showing an error? | Check input types and how the calculation handles missing values. Verify functions against current Fulcrum documentation. Do not silently replace missing information with zero unless that is the intended rule. |
Why does it work for some inputs but not others? | Compare these working and failing examples: [examples]. Check zero, blank, negative, and unexpected values where relevant. |
Why does the result not update as expected? | Check which fields the calculation depends on and help me reproduce the update. Distinguish the calculated value from how it is displayed and from any copied value stored elsewhere. |
Why are units, dates, or rounding wrong? | Check units, date handling, timezone where relevant, rounding rules, and display format. Explain the assumptions. Specify storage units separately from display units. |
Why is a score based on dropdown choices wrong? | Check how each selected choice maps to its score. Verify stored choice values rather than labels alone, and the handling of multiple selections if applicable. |
Why do a graphic and its measurement disagree? | Check the coordinate units and measurement definition: vertical height, horizontal setback, or full 3D distance. Recompute from the stored endpoints. A display change such as larger points should not change the measurement. |
How do I verify the fix? | Give me a small table of sample inputs and expected results, including relevant edge cases. For sensor or scan measurements, also define a physical test against a known reference; synthetic calculations alone do not prove real-world accuracy. |
Be specific about storage and display. For example: "Store duration in whole seconds and display minutes:seconds. Early Stop records elapsed time; completing the countdown records the configured duration; Cancel leaves the previous value unchanged."
Troubleshooting API integrations and external tools
Use these prompts for hosted pages, mobile companion apps, hardware, uploads, and services connected to Fulcrum. Describe the source, destination, trigger, and which system owns each value.
Useful keywords: REST API, native Workflow, webhook, authentication, field mapping, retries, duplicate prevention
Question or issue | Prompt to copy |
Which kind of integration should I use? | Compare native Workflows, webhooks, Data Events, and an external API service for [outcome]. Check current support and tool access. Identify where the action runs, what triggers it, and whether offline work changes when it can run. |
Why does the external tool get access denied? | Prove authentication and authorization for one real resource in the actual hosting or device environment. Check region, organization, and browser-origin restrictions. Do not assume launching a URL transfers Fulcrum login credentials. |
Why does a file upload fail? | Identify the failing stage: requesting upload details, transferring bytes, finalizing the upload, or linking it to the record. Capture the status and a redacted error body. Check expiry, required headers, file type, and size before retrying. |
Why are records or attachments duplicated or overwritten? | Inspect the latest destination state and the matching identifiers. Preserve existing attachments and unrelated fields. Make retries and repeated submissions safe, and explain how conflicting edits are handled. |
Why does it fail after a network interruption? | Check queue state, retry ownership, timeouts, and recovery. Show what succeeded and what remains pending. Test a duplicate submission and an interrupted transfer using approved test data. |
Is this demonstration doing real work? | List what is real, simulated, predefined, or unverified. Distinguish a displayed preview, browser-only storage, real API access, saved Fulcrum records, and automatic detection. Test one complete real path before expanding the feature. |
Why does it work locally but fail on my device? | Compare the target OS, device model, hardware/firmware, permissions, network, and file format. Provide redacted diagnostics and one focused physical test. Separate compilation and automated tests from device-tested behavior. |
How do I verify the fix? | Trace one item from source through transfer to the saved Fulcrum field, then retrieve or reopen it. Check success, interruption, duplicate submission, and recovery. Report any remaining hosting, deployment, or physical-device step. |
To learn more, see REST API authentication in the Fulcrum Developer Docs.
Help with any issue
Use these prompts to reset troubleshooting or prepare a clear escalation.
Useful keywords: reproduce, confirmed facts, hypothesis, smallest fix, Product feedback, support summary
Question or issue | Prompt to copy |
The AI says Fulcrum has a bug or limitation | Check current documentation, settings, permissions, missing files, and implementation errors before calling this a Fulcrum bug or limitation. Show the evidence and remaining uncertainty. |
The AI keeps trying fixes that do not work | Stop making changes. Summarize what you checked, what you tried, and the results. Revisit the hypothesis and propose one useful diagnostic test instead of another speculative fix. |
I am unsure what was actually completed | State what is saved in Fulcrum, local, unpublished, or untested. Distinguish schema validation, local tests, live execution, save/reopen checks, and physical-device verification. |
I need to continue in a new conversation | Summarize the goal, current resources, deployed version, decisions, findings, failed attempts, and next test. Omit secrets. Re-inspect live configuration before editing. |
I need to contact Support or Product | Prepare a summary with the affected resource, AI client, expected and actual behavior, reproduction steps, exact error, device/connectivity, and attempted fixes. Separate facts from guesses. For a capability request, explain the desired workflow and missing capability. Include relevant IDs, but no credentials. |
A reusable troubleshooting prompt
Fill in the brackets, then add the instruction at the end:
Target and environment: [resource; test/production; device/version; connectivity] Steps and evidence: [reproduction steps; redacted diagnostics] Expected / actual: [specific results and exact error] Last working / scope: [what changed; what may be edited] Inspect the current implementation. Separate facts from hypotheses, make the smallest justified fix, preserve unrelated behavior, and repeat the same test. State what passed and what still needs my interaction.
Contacting Support or Product
You don't need to finish every prompt in this article before asking for help.
Product feedback: For unexpected behavior in Fulcrum MCP or the AI Toolkit, or a capability you would like added, email [email protected].
Support: For data loss, blocked fieldwork, or widespread problems, contact Support directly through the chat in Fulcrum or by email at [email protected].
Notes
Fulcrum MCP and the AI Toolkit are Labs features, so behavior may change as we improve them.
The keywords in this article are routing cues, not guaranteed commands. The tools and permissions available to your AI assistant determine what it can do.
AI assistants can make mistakes. Test every change in an approved test context before you use it with live records.