Analyst Journeys
This document describes the complete workflows for analysts using Traces.
Journey 1: Creating a Connector
An analyst creates a new connector for a platform.
┌─────────────┐ ┌─────────────┐ ┌─────────────┐ ┌─────────────┐
│ Provide │ ──▶ │ Review │ ──▶ │ Config │ ──▶ │ Publish │
│ Doc URL │ │ Draft │ │ & Test │ │ │
└─────────────┘ └─────────────┘ └─────────────┘ └─────────────┘
Steps
-
Provide Documentation URL
- Navigate to Connectors → New
- Select "Generate from Documentation"
- Enter API documentation URL
- Provide platform name
- Start generation
-
Review Draft
- System parses and analyzes endpoints
- Review LLM recommendations
- Toggle endpoints enabled/disabled
- Add analyst notes
- Verify categories
-
Configure & Test
- In the connector editor, use the left tabs: Config (connector configuration, headers, URL), Bulk (bulk enable/disable endpoints), Auth (authentication), Test (test credentials), Notes
- For each endpoint, use the vertical tabs: Response, Config, Paging, Headers, AI Guide, Curl (plus History and Examine popouts)
- Configure pagination for list endpoints
- Set up dependency mappings
- Test individual endpoints, then the full dependency chain
- Group endpoints into execution Phases to control ordering
-
Publish
- Verify all configuration
- Publish creates a new connector version (new UUID) and soft-deletes the previous one for forensic integrity; endpoint references are remapped
- The connector is live for collections
Connector development happens in the draft editor (
/analyst/connectors/drafts/[id]); published connectors are edited at/analyst/connectors/[id]/edit.
Key Decisions
- Which endpoints to enable
- Pagination configuration
- Dependency order and execution phases
- Rate limit settings
Journey 2: Configuring Pagination
An analyst sets up pagination for an endpoint.
┌─────────────┐ ┌─────────────┐ ┌─────────────┐ ┌─────────────┐
│ Identify │ ──▶ │ Configure │ ──▶ │ Test │ ──▶ │ Verify │
│ Type │ │ Fields │ │ Pages │ │ Complete │
└─────────────┘ └─────────────┘ └─────────────┘ └─────────────┘
Steps
-
Identify Type
- Test endpoint to see response
- Look for pagination indicators:
- Cursor/token field
- Has_more flag
- Total count
- Link headers
- Determine pagination type
-
Configure Fields
- Enable pagination
- Select type (cursor, offset, page, link_header)
- Set parameter names
- Set response paths
- Configure stop condition
-
Test Pages
- Run test with pagination
- Verify second page requested
- Check cursor/offset increments
- Watch for stop condition
-
Verify Complete
- All pages retrieved
- Data count matches expected
- Stop condition triggered correctly
Journey 3: Setting Up Dependencies
An analyst maps dependencies between endpoints.
┌─────────────┐ ┌─────────────┐ ┌─────────────┐ ┌─────────────┐
│ Test │ ──▶ │ Select │ ──▶ │ Map │ ──▶ │ Test │
│ Source │ │ Path │ │ Target │ │ Chain │
└─────────────┘ └─────────────┘ └─────────────┘ └─────────────┘
Steps
-
Test Source
- Run test on source endpoint
- View response in JSON viewer
- Identify field with needed values
- Example:
channels[*].id
-
Select Path
- Click on field in JSON viewer
- Path is highlighted
- Or type path manually
-
Map Target
- Drag path to target endpoint
- Or use mapping dialog
- Set parameter name
- Set parameter location (path/query/body)
- Choose mode (iterate/first)
-
Test Chain
- Click "Test Dependency Chain"
- System runs endpoints in order
- Values extracted and substituted
- Review results
Journey 4: Reviewing a Collection
An analyst reviews and signs off on a completed collection.
┌─────────────┐ ┌─────────────┐ ┌─────────────┐ ┌─────────────┐
│ Check │ ──▶ │ Verify │ ──▶ │ Review │ ──▶ │ Sign │
│ Files │ │ Hashes │ │ Logs │ │ Off │
└─────────────┘ └─────────────┘ └─────────────┘ └─────────────┘
Steps
-
Check Files
- Open collection in review queue
- View file list
- Check file counts by endpoint
- Verify expected data present
-
Verify Hashes
- All files show hash status
- Confirm "Validated" for each
- Flag any validation failures
- Check hash algorithm used
-
Review Logs
- Check API logs for errors
- Review any failed endpoints
- Verify rate limits respected
- Note any anomalies
-
Sign Off
- Click "Sign Off"
- Add notes if needed
- Provide signature
- Collection moves to delivery
Review Checklist
- All enabled endpoints completed
- File counts reasonable
- Hashes validated
- No unexpected errors
- Chain of custody complete
Journey 5: Troubleshooting Failed Collection
An analyst diagnoses and resolves a failed collection.
┌─────────────┐ ┌─────────────┐ ┌─────────────┐ ┌─────────────┐
│ Check │ ──▶ │ Identify │ ──▶ │ Fix │ ──▶ │ Retry/ │
│ Error │ │ Cause │ │ Issue │ │ New │
└─────────────┘ └─────────────┘ └─────────────┘ └─────────────┘
Steps
-
Check Error
- Open failed collection
- Read error message
- Check which endpoint(s) failed
- Note error code
-
Identify Cause
- Review API logs
- Common causes:
- 401: Credential invalid/expired
- 403: Missing permissions
- 429: Rate limited
- 500: Platform error
- Check request/response details
-
Fix Issue
- Credential issue: Contact customer for new creds
- Connector issue: Update configuration
- Transient error: May resolve on retry
-
Retry or New Collection
- Transient errors: Retry failed endpoint
- Credential errors: Create new collection
- Connector errors: Fix connector, new collection
Journey 6: Reviewing Connector Version History
Each publish creates a new connector version. Analysts can review what changed.
- Open a connector and go to its History page
(
/analyst/connectors/[id]/history) — a dedicated git-style diff view keyed by endpoint method + path (it was previously a modal). - Soft-deleted prior versions are reconstructed for the diff via the version API.
The Connectors Workspace
The analyst connectors list (/analyst/connectors) surfaces two header panels to
help prioritize work:
- Recently Worked On — connectors you touched most recently.
- Up Next — queued items by priority.
A dedicated Roadmap page (/analyst/roadmap, linked from the nav) tracks
planned/in-progress connector work.
Navigation is a slim top bar plus a slide-out drawer (Dashboard, Collections, Connectors, Roadmap, Credentials, Customers) — not a fixed sidebar.
System Events
Collection State Transitions
pending
│
▼
validating_creds
│
├──▶ error (invalid credentials)
│
▼
queued
│
▼
in_progress
│
├──▶ paused (manual)
│ │
│ ▼
│ resumed
│ │
│ └──▶ in_progress
│
├──▶ error (collection failed)
│
├──▶ partial (some endpoints failed)
│
▼
completed
│
▼
reviewing
│
▼
signed_off
│
▼
delivered
Chain of Custody Events
| Event | Trigger | Data Recorded |
|---|---|---|
collection_created | Customer submits | Request details, customer ID |
credentials_validated | System validates | Validation result, timestamp |
collection_started | Worker begins | Start time, worker ID |
endpoint_completed | Each endpoint done | Endpoint ID, file count |
collection_completed | All endpoints done | Summary statistics |
analyst_assigned | Analyst takes review | Analyst ID |
analyst_reviewed | Review completed | Review notes |
analyst_signed_off | Sign-off submitted | Signature, timestamp |
zip_created | Package generated | File path, size, hash |
collection_delivered | Customer notified | Delivery method |
error_occurred | Any error | Error details, endpoint |
Notification Flow
| Event | Channel | Content |
|---|---|---|
| New collection | In-app | Assignment notice |
| Collection error | Error details | |
| Review required | In-app | Review queue update |
| Customer verified | In-app | New customer ready |