User guide
Meet Apiheron, make sense of your data and learn where to start. No technical background needed.
BEHIND YOUR APPLICATION
How much of the data reaching your app is actually used?
Imagine opening a product list. Your app asks a server for information; this is an API request. Apiheron observes that exchange: how long it took, how much data arrived and which fields the app’s code read. This helps your developer spot possible waste or delays.
- 01
You use your app
Once connected, open pages, try filters and explore details. A small tracking tool (SDK) in your browser records this activity.
- 02
Apiheron connects the observations
It compares request times, response sizes and field reads. Patterns worth investigating become “findings”.
- 03
Your team improves and checks again
Your developer reviews the evidence. After a code change, repeat the same actions and compare the new measurements.
Learn by exploring
The server sends 10 fields. Switch between the two views below to see how the app’s field reads and Apiheron’s interpretation change.
10 fields from the server
- Name
- Price
- Stock
- Description
- Category
- Supplier
- Weight
- Barcode
- Created
- Updated
✓ Checked: read by the code. Empty square: not read in this example.
Product list
Ceramic mug
€12
2 of 10 fields read
The list read only the name and price. The other 8 fields were not read in this interaction: a smaller response might be enough.
Remember: “not read” does not mean “delete”. Check the request and session counts behind real findings and exercise other screens and actions before deciding. A code read does not necessarily mean the field was displayed on screen.
First steps
A developer can handle the setup. Then use the app normally and review the results together.
- 1
Connect your app
An admin creates a project under Projects and allows the address where the app runs (its origin). A developer adds the setup code from the project page and connects axios if the app uses it.
- 2
Try a real task
Open your monitored app, visit a list, search, filter and open a record’s details. Data accumulates only for the flows you exercise.
- 3
Choose the project and time range
Select your project on the left. On data screens, use range in the top bar to choose a time period. Start at Overview, then open an endpoint from Findings.
- 4
Improve, then repeat the task
After your developer makes a change, repeat the same flow. Choose a time range covering the new attempt and review duration, size and findings, taking care not to mix in older measurements.
Viewers can read data. Developers can act on findings. Project, member and integration settings require an admin or owner.
Find the right screen
Each screen answers a different question. Select a card to open that page.
How are things overall?
Overview
Request counts, timing, errors and traffic over time. Your starting point.
What should we review first?
Findings
Potential issues grouped by importance and type. Open an endpoint to inspect why it was flagged.
Which data address needs attention?
Endpoints
Requests doing the same job grouped together. Explore Fields, Transforms and Requests in the detail view.
What happened in one request?
Requests
Each call’s address, status, duration and originating page when available. Useful for investigating one attempt.
How does data get connected?
Projects
Setup code, connection status, allowed addresses and collection keys live here.
How do we follow up?
Integrations
Optional AI advice and Papyro task cards. Once connected, use them in an endpoint’s Actions section.
In the workspace, Organization manages the team space and Members manages access roles. Audit log shows authorized members who changed what. Your account menu provides language, theme and personal settings.
Turn findings into action
A finding is a signal to investigate. Read the example and observation counts before deciding with your developer.
There may be excess data
A list needs only names, but receives hundreds of records or unread fields.
Next: check Fields and Transforms to see what was read or filtered.
The same work may be repeated
The same request repeats rapidly or the same data is stored under different cache keys.
Next: inspect request timing and source; distinguish development-only repeats.
Requests may be waiting for each other
Requests starting one after another can delay the screen. Timing alone does not prove a dependency.
Next: ask your developer to check whether the calls really need to run in sequence.
A response may contain a sensitive field
A field name such as password or api_key may have been observed. This does not mean its value was inspected.
Next: ask your developer whether this information needs to reach the browser.
How do importance and counts work?
Start with High, then review Medium. “Dev only” findings are specific to development and excluded from the main count. A developer can mute a finding or mark it as a false positive with a note; find it under Muted and restore it on the endpoint page.
One finding = one endpoint + one issue type. Even if it occurs 20 times, it counts as one finding; occurrences are shown separately. “No findings” only means no rule was triggered by the available observations.
Review a change from start to finish
Use Apiheron to collect evidence, investigate a finding and check the result after a change.
1. Choose a finding
Start with high-priority findings. Open its endpoint and inspect the recorded evidence before assigning work.
Open findings2. Review the evidence
Use the endpoint’s Fields section to see observed reads and their pages, scenarios and call sites. Missing or unknown tracking does not prove a field is unused.
Open endpoints3. Create a Papyro card
Use the endpoint’s Actions section to create a Papyro card when connected. Keep task assignment and delivery status in Papyro; Apiheron notes support the technical investigation.
Open findings4. Verify the fix
Open Verify fix on the same endpoint. Compare baseline and current releases using equivalent interactions. Without sufficient captures the screen explains what is missing. New high-severity findings and regressions appear as a compact overview summary.
Choose an endpoint
Advanced investigation and project settings
For complex issues, open advanced session investigation from Requests. Scenario recording is optional: give a task a name, record its interactions and repeat it after the change. The SDK release label identifies the build.
Inspect sessionsOptional browser console example in the monitored app (SDK required): run the first line, perform the task, then run the second line to finish.
window.apiheron.startScenario("checkout");
window.apiheron.endScenario();Projection and API/MCP controls are under advanced project settings. API/MCP is off by default and only an install Admin can enable it. These tools are optional; the regular workflow does not need them.
Quick glossary
Return here when an unfamiliar term appears. The ? icons on data screens also explain their metrics.
- API
- The way an app asks a server for information or an action. Think of a restaurant’s ordering channel.
- Endpoint
- The address for a particular job, such as /api/products for products. Calls to that address are examined together.
- Request and response
- “Get the products” is a request; the returned product information is the response. Reopening a page may create another request.
- Field
- One piece of information in a response: product name, price or stock. Apiheron tracks whether code reads it.
- Session
- A browser activity group used by the SDK. Session counts help you understand how many separate uses support an observation; they are not a person count.
- Duration · p50 · p95
- Duration is the time from a browser request starting until its response downloads. 1,000 ms = 1 second. Half of requests finish within p50; 95% finish within p95.
- Response size
- The uncompressed size of returned JSON. A larger package means more data; compression may make the actual network transfer smaller.
- Error rate
- The share of requests returning error status codes. An expired login or missing record can be a cause; check the status code in the details.
Common questions
Open a question for a short answer.
No data appears. Where should I start?
Check the selected project and widen the time range on a data screen. Exercise the monitored app. If it is still empty, open the project’s Connection status and Troubleshooting with your developer; the script, key, allowed origin or axios connection may need attention.
What information does Apiheron record?
It collects field names and paths, item counts, reads, timing and size instead of response values. Request addresses and query parameters may also appear in records. Data is stored on your Apiheron server. Optional AI or Papyro actions send the relevant summary to those services.
Does it fix problems automatically?
No. Apiheron measures and flags; your developer changes the code. With the integrations connected, you can request AI advice or create a Papyro task card. These actions require a developer role or higher.
Should we remove unread fields?
Do not remove them based only on this observation. A field may be read on another page, filter or action. Exercise enough flows, check request and session counts, then have your developer verify usage in code.
Why can’t I change some settings?
Permissions depend on your organization role. Viewers read, developers act on findings and admins manage projects, members and integrations. Ask your admin about role changes.
Your first round: one screen, one finding, one check.
Open Overview, choose a finding and inspect the observation behind it.