Before you begin
A few things to gather before starting. Everything here is free or has a free tier. None of this requires programming experience to set up.
What you need physically
Download the Audtheia repository first
The workflow templates, database schemas, and configuration files you need throughout this guide are all in the Audtheia GitHub repository. Download it now before moving on.
git clone https://github.com/AudtheiaOfficial/audtheia-environmental-monitoring.git
cd audtheia-environmental-monitoring
API keys and credentials you will need
You will collect these one by one as you work through each section. The table below is a complete reference so nothing gets missed.
| Platform | What you need | Where to find it | Used in |
|---|---|---|---|
| Roboflow | API key |
Settings → Roboflow API Key | Roboflow workflow, N8N |
| Anthropic | API key |
console.anthropic.com | Roboflow Python blocks |
| OpenAI (or any LLM you prefer) | API key |
platform.openai.com | All N8N AI agents (9 + 5) |
| IUCN Red List | API key |
apiv3.iucnredlist.org | N8N RTSP Analyst |
| API Ninjas | API key |
api-ninjas.com | N8N RTSP Analyst |
| Mapbox | Access token |
account.mapbox.com | N8N Cartography Mapper |
| PDFShift | API key |
pdfshift.io | N8N Daily Reporter |
| Airtable | Personal Access Token |
airtable.com/create/tokens | N8N workflows, dashboard |
| Airtable | Base ID (starts with app…) |
Your Airtable base URL | N8N workflows, dashboard |
| Airtable | Table IDs ×3 (starts with tbl…) |
Each table's URL in Airtable | N8N workflows, dashboard |
Several data sources (GBIF, iNaturalist, OpenStreetMap, Open-Meteo, OpenTopoData, Open Elevation, NOAA Co-OPS, SOILGRIDS) are completely free and require no account or API key. They work automatically once N8N is running.
Roboflow watches your video stream and identifies species in real time. When a detection happens, four pre-built code blocks inside the workflow handle the rest: drawing the live overlay on your video, converting the raw detection into structured data, running a quick AI environmental note, and sending the detection to your N8N analysis pipeline. You only need to enter your API keys, nothing else.
Create a free Roboflow account
Sign up at roboflow.com. The free tier covers everything Audtheia needs.
Choose a detection model for your species
Pick whichever option matches your field site:
Marine sponge research (start here)
Fork the official Porifera classifier workspace official-porifera-classifier-ju8er on Roboflow Universe. This is the reference implementation: 126 species, 16,915 training images, trained with RF-DETR Medium. Fork it into your own account and use it as-is, or fine-tune it further on your specific site's data.
Other ecosystems or species
Browse Roboflow Universe to find a pre-trained model for your target species. Fork a workspace and note the model endpoint string (format: workspace-slug/model-id/version).
Train your own custom model
Upload and label your own images in Roboflow, then train a model using Roboflow's training tools. This gives you a model tailored to your exact species and environmental conditions.
Import the Roboflow Workflow template
Find the workflow template file in the roboflow-workflows/ folder of the repository you cloned. In your Roboflow workspace, go to Workflows and import the file. The workflow comes with five pre-configured components already in place:
- Object detection block runs your chosen model against each video frame and produces detection results
- Detection_Converter converts the raw detection output into clean structured data
- Add_Webcam_Interface generates the live species identification overlay on your video feed
- Anthropic_Environmental_Analyzer sends detection data to an AI language model at regular intervals for a quick environmental note
- Analyst_Caller forwards the structured detection to your N8N analysis workflow via a web address (webhook)
Repository location: roboflow-workflows/
All the logic is already built into the template. You only need to fill in your API keys. No coding required.
Add your API keys to the workflow blocks
Open each Python block in the workflow and enter the values listed below:
- Anthropic_Environmental_Analyzer: set
ANTHROPIC_API_KEYto your Anthropic API key - Analyst_Caller: set
webhook_urlto the N8N webhook URL you will get in Section 02, Step 5
You need the N8N webhook URL from Section 02 before you can finish this step. Complete Section 02 Steps 1 through 5, then come back here.
Note down your Roboflow API key and model endpoint
Go to Settings → Roboflow API Key and copy your key. Also note the model endpoint string from your model's deployment page. Both values are needed when you set up N8N in Section 02.
Connect your video source
Point the video input in the workflow at your camera or recording. Supported formats:
- RTSP stream from an IP camera (e.g.,
rtsp://192.168.1.100:554/stream) - An MP4 or other video file for testing
- A USB webcam connected to your computer
If your camera doesn't automatically transmit GPS coordinates, note your site's latitude and longitude for manual entry in N8N (Section 02, Step 7).
Run the workflow and verify detections appear
Start the workflow and confirm that species detections show up in the Roboflow output panel. Full end-to-end testing (confirming detections arrive in Airtable) happens after you complete Section 02.
Roboflow documentation
Every time Roboflow detects a species, it sends a message to N8N. Nine specialized AI agents then go to work: one identifies the full taxonomy, another checks whether this individual has been seen before, a third validates the GPS location, others look up the conservation status, pull real-time environmental conditions, assess the habitat, and generate a satellite map. The entire output, 72 structured data points, gets written to your Airtable database.
Create an N8N account or host it yourself
Sign up at n8n.io for a cloud account (free trial available), or deploy N8N on your own server. Both work with the Audtheia workflow template.
Import the RTSP Analyst workflow
Find RTSP_Analyst_Workflow__GitHub_Template_.json in the n8n-workflows/rtsp-analyst/ folder of the repository. In N8N, go to Workflows → Import and upload the file.
Repository location: n8n-workflows/rtsp-analyst/
Add your credentials to N8N
Go to Settings → Credentials → Add Credential and add the following. These are stored securely inside N8N and never leave your instance:
| Credential | Type | Used by |
|---|---|---|
| OpenAI API (or compatible LLM) | API key (Authorization: Bearer) | All 9 AI agents |
| Airtable | Personal Access Token | All Airtable write nodes, Memory Manager agent |
| IUCN Red List | API key (request header) | Biodiversity Intelligence agent |
| API Ninjas | API key (request header: X-Api-Key) |
Systematics Phenologist, Biodiversity Intelligence agents |
| Mapbox | Access token (query parameter: access_token) |
Cartography Mapper agent |
GBIF, iNaturalist, OpenStreetMap Nominatim, all Open-Meteo endpoints, OpenTopoData, Open Elevation, NOAA Co-OPS, and SOILGRIDS are completely free and require no credentials. They work automatically.
Enter your Airtable database identifiers
In the workflow's configuration nodes, enter your Airtable Base ID and Table IDs. You can also set them as environment variables:
AIRTABLE_BASE_ID=appXXXXXXXXXXXXXX
AIRTABLE_SPECIES_TABLE_ID=tblXXXXXXXXXXXXXX
AIRTABLE_MAPPING_TABLE_ID=tblXXXXXXXXXXXXXX
IUCN_API_KEY=your_iucn_key
API_NINJAS_KEY=your_api_ninjas_key
MAPBOX_ACCESS_TOKEN=your_mapbox_token
You will get the Base ID and Table IDs when you set up Airtable in Section 04. Come back here to fill these in after that.
Copy the webhook URL
Open the Webhook trigger node in the workflow and copy the URL it shows. It will look something like this:
https://your-n8n-instance.app.n8n.cloud/webhook/rtsp-analyst
Go back to Section 01, Step 4 and paste this URL into the webhook_url field in the Roboflow Analyst_Caller block.
Choose which climate agents to enable for your habitat
The RTSP Analyst has separate agents for marine and terrestrial environments. Enable only the ones that match your study site to keep things efficient:
| Your study environment | Marine Climate Search | Terrestrial Climate Search | Cartography Mapper |
|---|---|---|---|
| Marine | Enable | Disable | Enable when satellite imagery needed |
| Terrestrial | Disable | Enable | Enable when satellite imagery needed |
| Estuarine or mixed | Enable | Enable | Enable when satellite imagery needed |
The Cartography Mapper generates Mapbox satellite images at three zoom levels and uses a vision-capable AI model to interpret them. It is optional. Enable it only when spatial satellite imagery is genuinely useful for your research.
Set your site location (if your camera doesn't transmit GPS)
If your camera or recording device doesn't automatically include GPS coordinates, open the Set Static Location node on the left side of the workflow canvas and enter your site's latitude and longitude. All observations will use these coordinates.
Send a test detection to check everything is connected
Run this command in your terminal to send a test detection through the workflow:
curl -X POST https://your-n8n-instance.app.n8n.cloud/webhook/rtsp-analyst \
-H "Content-Type: application/json" \
-d '{
"species_name": "Test Species",
"scientific_name": "Genus species",
"confidence": 0.94,
"timestamp": "2025-01-01T14:32:15Z",
"location": {
"latitude": 10.234,
"longitude": -84.567
}
}'
A successful response looks like: {"success": true, "observation_id": "...", "data_points": 72, "agents_executed": 9}
Then check your Airtable database. A new record should appear in the Species Observations table. If the Cartography Mapper is enabled, a record should also appear in the Environmental Mapping table.
Activate the workflow
Toggle the Active switch in the top-right corner of the N8N workflow editor. The workflow will now receive live detections from Roboflow automatically. Check the Executions tab at any time to see what's been processed and any error details.
What the nine agents do
| Agent | Data sources | What it records |
|---|---|---|
| Systematics Phenologist | GBIF, iNaturalist, API Ninjas | Full taxonomy, seasonal timing, behavioral patterns (15 data points) |
| Memory Manager | Your Airtable records | Previous sightings of this individual, pattern changes, anomalies (5 data points) |
| GIS Data Manager | OpenStreetMap, OpenTopoData, Open Elevation | Confirmed GPS location, elevation, biogeographic region (8 data points) |
| Biodiversity Intelligence | GBIF, IUCN Red List, iNaturalist, API Ninjas | Conservation status, rarity score, ecological role (9 data points) |
| EnviroStatus Manager | Open-Meteo (4 APIs) | Habitat type, ecosystem classification, environmental suitability (6 data points) |
| Marine Climate Search | NOAA Co-OPS, Open-Meteo Marine, Open Elevation | Sea surface temperature, currents, wave height, salinity (8 data points, marine only) |
| Terrestrial Climate Search | Open-Meteo Weather, SOILGRIDS, Open Elevation | Temperature, rainfall, humidity, soil type, biome (8 data points, terrestrial only) |
| Habitat Assessment | NDVI, Open Elevation, Open-Meteo | Vegetation index, ecosystem health, habitat quality, degradation indicators (7 data points) |
| Cartography Mapper | Mapbox Static Images API, NOAA | Satellite imagery at three zoom levels, bathymetric data (10 data points, optional) |
Full documentation: n8n-workflows/rtsp-analyst/README.md
The Daily Reporter runs automatically on a schedule. Five agents read all the observations from the reporting period and each takes on a different analytical section of the report: species diversity, environmental and climate patterns, habitat health, satellite mapping, and a full FAIR-compliant summary. The finished PDF is saved to your Airtable Daily Reports table and downloadable from the researcher dashboard.
Import the Daily Reporter workflow
Find Daily_Reporter__GitHub_Template_.json in the n8n-workflows/daily-reporter/ folder. In N8N, go to Workflows → Import and upload the file.
Repository location: n8n-workflows/daily-reporter/
Add or reuse credentials
This workflow uses three credentials. If you already added the AI model and Airtable credentials in Section 02, you can reuse them here:
| Credential | Type | Used by |
|---|---|---|
| OpenAI API (or compatible LLM) | API key | All 5 AI agents |
| Airtable | Personal Access Token | Data retrieval and report saving |
| PDFShift | API key | PDF generation |
Get your PDFShift API key at pdfshift.io. The free tier allows up to 50 PDF conversions per month.
PDFShift is required for report generation. Without a valid PDFShift key, the workflow will complete its analysis but no PDF will be saved and no Airtable record will be created.
Enter your Airtable identifiers
Configure the four Airtable values this workflow needs:
AIRTABLE_BASE_ID=appXXXXXXXXXXXXXX
AIRTABLE_SPECIES_TABLE_ID=tblXXXXXXXXXXXXXX
AIRTABLE_MAPPING_TABLE_ID=tblXXXXXXXXXXXXXX
AIRTABLE_REPORTS_TABLE_ID=tblXXXXXXXXXXXXXX
Set the report schedule
Open the Daily Trigger node and enter a cron expression for how often you want reports generated. The default is midnight UTC every day:
0 0 * * * # Every day at midnight UTC (default)
0 6 * * * # Every day at 06:00 UTC
0 */6 * * * # Every 6 hours
0 0 * * 1 # Every Monday at midnight UTC
Run a manual test
Click Execute Workflow to trigger a report now. Check the Executions tab as it runs. When finished, open your Airtable Daily Reports table and confirm a new record appeared with a PDF URL. Reports typically take 2 to 5 minutes depending on how many observations exist.
If no observations exist yet for the reporting period, the workflow produces an empty-period notification rather than a full PDF report. Make sure the RTSP Analyst has processed at least one observation first.
Activate the workflow
Toggle the Active switch in the workflow editor. Reports will now generate automatically on your chosen schedule.
What the five agents write in each report
| Agent | Report section |
|---|---|
| Agent 1: Species and Biodiversity Analyst | Species diversity indices, conservation status breakdown, rare and endemic species, ecological roles, activity patterns over the period |
| Agent 2: Environmental and Climate Analyst | Climate trends, sea surface temperature, precipitation, atmospheric conditions, oceanographic patterns, seasonal context, anomaly flags |
| Agent 3: Habitat and Conservation Analyst | Habitat type distribution, ecosystem health scores, threat indicators, degradation, management priority recommendations |
| Agent 4: Cartography and Spatial Analyst (uses vision) | Interprets satellite imagery across three zoom levels, geographic clustering, habitat connectivity, bathymetric analysis, land cover |
| Agent 5: Report Synthesis and Generation | Combines all sections into a structured narrative, assembles a 12-section HTML report, documents FAIR compliance, generates the final PDF via PDFShift |
Full documentation: n8n-workflows/daily-reporter/README.md
Airtable is the database that holds all of your observation data. The N8N workflows write to it and the Audtheia dashboard reads from it. You need to create three tables. The fastest way is to import the example CSV files from the repository, which creates the right column structure automatically. Column names must match exactly as listed, because the N8N workflows reference them by name.
Schema documentation and example CSV files are in: airtable-schemas/
Create a free Airtable account
Sign up at airtable.com. The free tier fully supports the Audtheia database schema.
Create a new base named "Environmental Data Management"
This exact name is referenced during Personal Access Token setup. Inside this base you will create the three tables below.
Create the Species Observations table (62 columns)
Create a table named Species Observations. The easiest method is to import species-observations-example.csv from airtable-schemas/, which sets up the columns automatically. Then adjust each column's data type to match the definitions below.
Schema files and example CSVs: airtable-schemas/
Taxonomic and Identification — Columns 1 to 5
| Column name | Type | What it stores |
|---|---|---|
| Species Name | Short Text | The common name of the observed species |
| Species Taxonomy | Short Text | Full classification from Kingdom to Species, separated by semicolons |
| Observation ID | Short Text | Unique identifier for this observation (e.g., OBS-12345) |
| Observation Time | Number | Unix timestamp of when the detection occurred |
| Total Species Count | Number | Number of individuals visible in the detection frame |
Spatiotemporal Context — Columns 6 to 7
| Column name | Type | What it stores |
|---|---|---|
| Same Specimen | Checkbox | Whether this individual has been detected before |
| Observation Season | Single Select | Winter, Spring, Summer, or Autumn |
Geographic Data — Columns 8 to 12
| Column name | Type | What it stores |
|---|---|---|
| Location | Short Text | Specific place name, reverse-geocoded from GPS coordinates |
| Region | Short Text | Geographic region or state |
| Country | Short Text | Country name |
| Geographic Confidence | Single Select | High, Medium, Low, or Unknown |
| GIS Analysis | Long Text | Detailed geographic analysis: biogeographic province, protected area proximity, habitat fragmentation, human footprint index |
Environmental Conditions — Columns 13 to 15
| Column name | Type | What it stores |
|---|---|---|
| Environmental Conditions | Long Text | Overall environmental assessment at the time of observation |
| Weather Conditions | Long Text | Real-time weather readings for the site |
| Habitat Type Classification | Single Select | Marine, Freshwater, Estuarine, Terrestrial, Mixed, or Unknown |
Geographic and Marine Zone — Columns 16 to 17
| Column name | Type | What it stores |
|---|---|---|
| Geographic Zone | Short Text | Biogeographic zone (e.g., "Tropical Marine Zone (MEOW Province #11)") |
| Marine Zone | Short Text | Marine habitat type (e.g., "Neritic Coral Reef", "Intertidal Rocky Shore") |
Marine Climate Data — Columns 18 to 23
| Column name | Type | What it stores |
|---|---|---|
| Air-Sea Interface (SST, Salinity, Wind) | Long Text | Sea surface temperature, salinity in PSU, wind speed and direction |
| Atmospheric Pressure | Number | Pressure in hPa |
| Ocean Current Velocity-Direction | Long Text | Current speed in m/s and cardinal direction |
| Substrate Type | Single Select | Sand, Rock, Coral, Mud, Soil, Gravel, Seagrass, Algae, Mixed, or Unknown |
| Water Depth | Number | Depth in meters |
| Wave Height | Number | Significant wave height in meters |
Terrestrial Climate Data — Columns 24 to 32
| Column name | Type | What it stores |
|---|---|---|
| Terrestrial Biome | Short Text | Biome type (e.g., "Tropical Rainforest", "Temperate Grassland") |
| Air Temperature (Celsius) | Number | Air temperature in degrees Celsius |
| Soil Temperature (Celsius) | Number | Soil temperature in degrees Celsius |
| Precipitation (mm) | Number | Precipitation in millimetres |
| Humidity % | Number | Relative humidity, 0 to 100 |
| Elevation (Meters) | Number | Elevation above sea level in metres |
| Land Cover Type | Single Select | Snow/Ice, Water, Wetland, Grassland, Shrubland, Barren, Cropland, Urban, Forest, or Unknown |
| Soil Type | Single Select | Clay, Sand, Loam, Silt, Peat, Rocky, Mixed, or Unknown |
| Vegetation Index | Number | NDVI score, 0.0 to 1.0 |
Chronobiological Data — Columns 33 to 34
| Column name | Type | What it stores |
|---|---|---|
| Chronobiological Observations | Multiple Select | Activity period (Diurnal, Nocturnal, Crepuscular, Unknown), behavioral context (Feeding, Resting, Sessile, Social, Reproductive, Territorial, Migration, Invasive), circadian phase (Activity Onset, Activity Offset, Acrophase, Arrhythmic) |
| Chronobiology Notes | Long Text | Narrative analysis of activity timing, circadian patterns, and behavioral context |
Phenological Data — Columns 35 to 36
| Column name | Type | What it stores |
|---|---|---|
| Phenological Observations | Multiple Select | Seasonal timing (Expected, Early, Late, Unusual, Unknown), life cycle stage (Juvenile, Adult, Reproductive, Migrating, Dormant, Unknown), breeding season (Pre-breeding, Breeding, Post-breeding, Non-breeding, Unknown) |
| Phenology Notes | Long Text | Narrative assessment of seasonal timing, life stage, breeding season, and climate correlations |
Biodiversity Data — Columns 37 to 45
| Column name | Type | What it stores |
|---|---|---|
| GBIF Occurrence Count | Number | Total recorded occurrences for this species in GBIF globally |
| IUCN Conservation Status | Single Select | LC, NT, VU, EN, CR, EW, EX, DD, or NE |
| Species Abundance Trend | Single Select | Increasing, Stable, Decreasing, Fluctuating, or Unknown |
| Species Rarity Score | Number | 0.0 (common) to 1.0 (extremely rare), based on GBIF occurrence data |
| Species Rarity Reason | Long Text | Explanation of the rarity score: distribution, occurrence data, and regional threats |
| Endemic Species | Checkbox | Whether this species is endemic to the observed region |
| Breeding Season Match | Checkbox | Whether this observation falls within the species' known breeding season |
| Migration Route Overlap | Checkbox | Whether this location overlaps a known migration route |
| Ecological Role | Long Text | The species' function in the ecosystem (e.g., "Pollinator", "Filter feeder", "Apex predator") |
Conservation and Threat Assessment — Columns 46 to 52
| Column name | Type | What it stores |
|---|---|---|
| Environmental Conservation Status | Long Text | Protected area status, IUCN category, habitat coverage, protection level |
| Threat Indicators | Long Text | Identified threats: tourism pressure, development, pollution, climate impacts |
| Pollution Risk Level | Single Select | None, Low, Moderate, High, Severe, or Unknown |
| Algal Bloom Risk | Single Select | None, Low, Moderate, High, Active Bloom, or Unknown |
| Overfishing Pressure | Single Select | None, Low, Moderate, High, Overexploited, or Unknown |
| Invasive Species Risk | Single Select | None, Low, Moderate, High, Established, or Unknown |
| Deforestation Risk | Single Select | None, Low, Moderate, High, Active Clearing, or Unknown |
Habitat Assessment — Columns 53 to 60
| Column name | Type | What it stores |
|---|---|---|
| Canopy Cover Percent | Number | Forest canopy coverage percentage, 0 to 100 |
| Habitat Complexity | Number | Structural complexity score, 0 to 100 |
| Habitat Connectivity | Number | Connectivity index, 0 to 100 |
| Anthropogenic Disturbance | Single Select | None, Low, Moderate, High, Severe, or Unknown |
| Geopolitical Stability | Single Select | Stable, Generally Stable, Moderate Risk, High Risk, Conflict Zone, or Unknown |
| Habitat Quality Score | Number | Overall habitat quality, 0 to 100 |
| Ecosystem Health Score | Number | Composite ecosystem health metric, 0 to 100, weighted from NDVI, water quality, pollution, and biodiversity |
| Conservation Timing Priority | Single Select | Low, Medium, High, Critical, or Unknown |
Research Metadata — Columns 61 to 62
| Column name | Type | What it stores |
|---|---|---|
| Research Value | Number | Scientific research value score, 0 to 100 |
| Monitoring Frequency | Single Select | Daily, Weekly, Monthly, Quarterly, Annually, As Needed, or Unknown |
Create the Environmental Mapping table (10 columns)
Create a table named Environmental Mapping with exactly these 10 columns. Column names must match precisely:
| Column name | Type | What it stores |
|---|---|---|
| Species Name | Short Text | Species observed at this location |
| Observation ID | Short Text | Links to the Species Observations record by Observation ID |
| Map Overview URL | URL | Mapbox satellite image at zoom level 10 |
| Map Detail URL | URL | Mapbox satellite image at zoom level 12 |
| Map Closeup URL | URL | Mapbox satellite image at zoom level 14 |
| Center Coordinates | Long Text | Latitude, longitude of the detection site |
| Bathymetry Available | Checkbox | Whether NOAA bathymetric data is available for this location |
| Research Stations Marked | Checkbox | Whether NOAA research stations appear on the generated map |
| Map Generation Date | Date | Timestamp of when the Mapbox images were generated |
| Projection System | Short Text | Map projection used (e.g., "Web Mercator", "WGS 84") |
Create the Daily Reports table (3 columns)
Create a table named Daily Reports with these 3 columns:
| Column name | Type | What it stores |
|---|---|---|
| Report Name | Short Text | Primary field. Report identifier (e.g., "Audtheia Daily Environmental Report - 2025-11-27") |
| Report PDF URL | URL | Download link for the PDF generated by PDFShift |
| Date Uploaded | Date | When the report was generated |
Create a Personal Access Token
Go to Builder Hub → Personal access tokens and create a new token. Before saving, configure the following:
- Scopes: add
data.records:read(for the dashboard) anddata.records:write(for N8N to write observations) - Access: add your Environmental Data Management base to the token's access list
Copy the token immediately after creation. Airtable shows it only once. If you lose it, you must generate a new one and update all N8N workflows and the dashboard.
A missing data.records:read scope is the most common cause of "Invalid permissions" errors in the dashboard. Adding the base to the token's access list is also required, even when the scopes are correct.
Note down your Base ID and Table IDs
These appear in the Airtable URL when you open your base and each table. You need one Base ID and three Table IDs:
# Base ID is in the base URL (starts with "app")
https://airtable.com/appXXXXXXXXXXXXXX/...
^^^^^^^^^^^^^^^^
# Table ID appears when a specific table is open (starts with "tbl")
https://airtable.com/appXXXXXXXXXXXXXX/tblXXXXXXXXXXXXXX/...
^^^^^^^^^^^^^^^^
Record these four values. Enter them into N8N (Sections 02 and 03) and into the dashboard settings (Section 05).
The dashboard reads directly from your Airtable base using your own credentials. You enter your token and base identifiers once in the dashboard settings, and they are saved locally in your browser. Nothing is sent to Audtheia. No Audtheia account is needed.
Open the dashboard
Go to the Audtheia dashboard. The settings panel appears automatically on the first visit.
Enter your Airtable credentials in Settings
Fill in all five fields:
- Airtable Personal Access Token: the token you created in Section 04, Step 6
- Airtable Base ID: starts with
app - Species Observations Table ID: starts with
tbl - Environmental Mapping Table ID: starts with
tbl - Daily Reports Table ID: starts with
tbl
Save and verify the connection
Click Save Credentials. The dashboard will attempt to connect to your Airtable base. If credentials are correct and observations exist, the Species Observations tab will populate with records. If the table is empty, that is expected until the RTSP Analyst has processed at least one detection.
Troubleshooting
If something isn't working, these are the most common causes. Check the N8N Executions tab first, it shows exactly where a failure occurred and what the error message was.
- Confirm the N8N RTSP Analyst workflow is Active (toggle is on in the top-right corner of the workflow editor).
- Confirm the webhook URL in the Roboflow
Analyst_CallerPython block matches exactly what the N8N Webhook trigger node shows. - Check that the Roboflow workflow is running against a live video source and producing detections in the output panel.
- Confirm the Personal Access Token includes both
data.records:readanddata.records:writescopes. - Confirm the Environmental Data Management base was added to the token's access list when you created it. This is a separate step from adding scopes and is easy to miss.
- Check that the Base ID starts with
appand all three Table IDs start withtbl. Copying the wrong URL segment is the most common mistake here.
- Confirm the Cartography Mapper agent is enabled in the RTSP Analyst workflow (the toggle in the agent node).
- Verify your Mapbox access token is correctly entered in N8N credentials. An invalid token fails silently and produces empty URL fields in the Airtable record.
- Check your Mapbox account at account.mapbox.com for token status and monthly usage quota.
If credentials are accepted without an error but tabs show nothing:
- Verify the Base ID starts with
appand each Table ID starts withtbl. - Confirm the RTSP Analyst workflow has processed at least one observation. Tables must contain records before the dashboard has anything to show.
- Check that table names in Airtable match exactly: Species Observations, Environmental Mapping, Daily Reports, including capitalization and spacing.
- Verify the PDFShift API key is correctly entered in N8N credentials under Settings → Credentials.
- Confirm observations exist for the reporting period. Run the RTSP Analyst first and verify records appear in Airtable before triggering the Daily Reporter.
- Verify the Daily Reports table exists with column names exactly as specified in Section 04, Step 5: Report Name, Report PDF URL, Date Uploaded.
- Check the N8N Executions tab for the Daily Reporter workflow to find specific node errors.
Partial records mean one or more N8N agents failed mid-run. Check the N8N Executions tab for errors on specific agent nodes.
- IUCN or API Ninjas key missing or rate-limited: verify credentials and account quota.
- Agent execution timeout: increase the timeout setting in N8N Settings. Reducing
max_tokensin the agent configuration can also help. - Airtable Personal Access Token missing the
data.records:writescope: verify in Builder Hub. - Column name mismatch in Airtable: confirm all 62 Species Observations column names match the schema exactly, including capitalization and spaces.
This is normal. The Audtheia demo proxy runs on Render.com's free tier, which pauses the service after 15 minutes without traffic. The first request wakes it up, which takes 30 to 50 seconds. After that, requests are fast. This only affects the public demo at demo.html and has no impact on your own Roboflow or N8N deployment.
Getting Help
If you run into something not covered above, open a GitHub issue. Include the N8N execution log, the error message from Airtable or the dashboard, and the section and step number where things stopped working. This helps narrow it down quickly.
The repository also has full workflow documentation, Airtable schema files, and annotated example CSV files that cover edge cases not addressed in this guide.