Getting Help
Get instant answers about Trinity through the AI-powered documentation assistant, or connect with the community.
Trinity Docs Q&A Bot
Ask questions about Trinity and get grounded answers from the documentation.
CLI Usage
./scripts/ask-trinity.sh "How do I create an agent?"
./scripts/ask-trinity.sh "What credentials do agents need?"
./scripts/ask-trinity.sh "How do I troubleshoot a stuck agent?"API Usage
curl -X POST \
-H "Content-Type: application/json" \
"https://us-central1-mcp-server-project-455215.cloudfunctions.net/ask-trinity" \
-d '{"question": "How do I schedule an agent task?"}'Response:
{
"answer": "To schedule an agent task...",
"state": "SUCCEEDED"
}No authentication required. The bot uses Vertex AI Search with Gemini to generate answers from the onboarding documentation.
Report a Bug, Request a Feature, or Send Feedback
A floating Help widget lets you file a report without leaving Trinity. It has four tabs:
How it works
Pick the Bug, Feature, or Feedback tab.
Enter a title and description. Optionally leave a contact email if you want a reply.
Review the see-before-send screen, which shows exactly what will be submitted.
Confirm to send. Nothing is submitted until you confirm.
Where reports go
abilityai/trinity. The success screen links to the created issue.What's captured
Each report includes a small diagnostics bundle to speed up triage: app version and git commit, the current page, browser user agent, viewport, OS, and the last few console errors or warnings. Tokens, credentials, emails, and private IP addresses are automatically scrubbed — both before you review and again on the server — so secrets aren't exposed in a public issue. You see the already-scrubbed payload on the review screen.
Your self-hosted instance never stores a GitHub token; reports post to an Ability.ai-operated intake endpoint that files the issue on your behalf. Operators can disable the report tabs or repoint the intake URL at build time.