Skip to main content

Automated Connector Generation

The recommended way to create connectors is through automated generation. The system parses API documentation and uses AI to analyze endpoints for forensic relevance.

Supported Documentation Formats​

FormatDescription
OpenAPI 3.xModern API specification (preferred)
Swagger 2.0Legacy OpenAPI format
Postman CollectionExported from Postman
HTML DocumentationParsed with AI assistance

Starting Generation​

  1. Navigate to Connectors → Generate from API Docs
  2. Choose your input:
    • URL tab: paste a link to an OpenAPI spec, Swagger file, or Postman collection
    • Upload File tab: upload a .json, .yaml, or .yml spec file
  3. Click Generate Connector

There is no separate platform-name or base-URL field — those are derived from the specification. After you submit, a progress screen tracks parsing and analysis; when it finishes, click Review Draft to open the draft editor.

Generation Process​

Phase 1: Parsing​

The system:

  1. Fetches the documentation URL
  2. Detects the format
  3. Extracts endpoint definitions
  4. Identifies authentication requirements

Phase 2: Analysis​

AI analyzes each endpoint for:

  • Forensic relevance - Evidence value
  • Data modification - Read-only preference
  • Category - Logical grouping
  • Friendly name - Human-readable description
  • Keep/discard recommendation - With reasoning

Phase 3: Draft Creation​

The system creates a draft with:

  • All parsed endpoints
  • AI recommendations
  • Suggested pagination configurations
  • Dependency hints

Reviewing the Draft​

Review happens in the draft editor (the "Review: {platform}" page), which has three tabs:

Review Endpoints Tab​

View all endpoints with:

  • Filter by: enabled/disabled, category, search
  • LLM recommendations with reasoning
  • Toggle endpoints enabled/disabled
  • Add analyst notes
  • Stats: total endpoints, enabled count, categories

Endpoint Explorer Tab​

A full-height explorer for browsing and selecting endpoints alongside the endpoint editor.

Test & Configure Tab​

  • Add test credentials
  • Test individual endpoints
  • View response data
  • Configure pagination
  • Set up dependency mappings

Draft Review Actions​

ActionDescription
Toggle enabledEnable or disable an endpoint
View recommendationSee AI reasoning for keep/discard
Add notesDocument analyst decisions
Override categoryChange auto-assigned category
Mark forensically relevantOverride relevance assessment

What the AI Analyzes​

Forensic Relevance​

AI looks for endpoints that return:

  • User profiles and identities
  • Messages and communications
  • Files and documents
  • Activity logs and audit trails
  • Timestamps and metadata

Modification Detection​

AI identifies endpoints that:

  • Create new resources (POST)
  • Update existing resources (PUT/PATCH)
  • Delete resources (DELETE)
  • Trigger actions or workflows

Category Assignment​

Common categories:

  • users - User accounts and profiles
  • messages - Communications
  • files - Documents and attachments
  • channels - Groups and conversations
  • activity - Logs and audit trails
  • metadata - Configuration and settings

Finalizing the Connector​

Once you've reviewed all endpoints:

  1. Ensure at least one endpoint is enabled (the Create Connector button is disabled otherwise)
  2. Test critical endpoints with real credentials
  3. Verify pagination works correctly
  4. Click Create Connector

The connector is finalized at version 1.0.0 with Testing status automatically — there's no version to enter.

Troubleshooting Generation​

Generation Fails During Parsing​

Possible causes:

  • URL not accessible
  • Unsupported format
  • Invalid specification

Solutions:

  1. Verify URL is publicly accessible
  2. Try different documentation URL
  3. Check for valid OpenAPI/Swagger spec

Missing Endpoints​

Possible causes:

  • Endpoints not in documentation
  • Parsing error
  • Hidden or internal endpoints

Solutions:

  1. Add missing endpoints manually
  2. Try different documentation source
  3. Review parsed endpoints carefully

Incorrect Recommendations​

The AI may incorrectly classify endpoints. Always:

  1. Review all recommendations
  2. Override as needed
  3. Add analyst notes explaining changes

Next Steps​