Tickety OPS Tower

Self-hosted intelligence for IT service operations.

Tickety brings ticket queues, SLA monitoring, reports, knowledge, business requirements, and optional AI analysis into one workspace. In production, it reads Freshservice records into a local store without writing ticket changes back to Freshservice.

Built with FastAPI, Next.js, and PostgreSQL. Runs with Docker Compose or Kubernetes/Helm.

Requirements workspace

Open Work → Requirements (/requirements) to capture an approved project, request, or enhancement around a business objective. Evidence, questions, requirements, and user stories stay connected in one workspace. Worth your attention suggests the next useful action from current blockers and business priorities; users can revisit any part of the work without following a fixed wizard.

• Bring business context: paste material or preview TXT, Markdown, EML, DOCX, selectable-text PDF, VTT, and SRT files. Classify sources as documents, emails, SOPs, or meeting transcripts. Saved evidence is immutable and traceable.
• Clarify scope: capture exact evidence excerpts, outcomes, priorities, and observable acceptance criteria. Assign business questions to an answer-owner role and distinguish blocking decisions from exploratory questions. Keep background material and deferred needs without treating them as current delivery.
• Review with accountability: compare selected requirements for possible conflicts using optional AI assistance. Humans record decisions and sign off a specific revision with their review capacity and rationale. Changes to agreed scope withdraw its sign-off and story until reviewed again.
• Prepare the handoff: create user stories from signed-off requirements, copy their acceptance criteria and provenance, or export a Markdown BRD with priorities, evidence, decisions, and outstanding exploration. These actions do not create external development tickets or constitute deployment approval.

An active session is required, including in demo mode. Initiatives are accessible to their creator and active administrators. AI requires a configured, available provider; manual work remains available without it. Suggestions never sign off requirements or automatically record business decisions.

Uploads are limited to 400 KB and confirmed source text to 100,000 characters. Scanned PDFs require OCR or transcription beforehand. Unsaved drafts stay only in the current tab's memory while navigating: save before refreshing or signing out. See the business guide and runtime boundaries for details.

Quick start

For the maintained dev branch on MicroK8s, use the Dev deployment guide. The isolated Compose evaluation path below is a separate deployment option.

Requires Docker and Docker Compose 2.24 or later. Run these commands from the repository root.

1. Copy the environment template:

   cp .env.example .env

2. Edit .env for isolated local evaluation:

   APP_MODE=demo
   LOGIN_REQUIRED=true
   ITSM_PROVIDER=standalone
   FRONTEND_URL=http://localhost:3000
   CORS_ALLOW_ORIGINS=http://localhost:3000
   POSTGRES_PASSWORD=replace-with-a-long-random-password
   DATABASE_URL=postgresql+psycopg2://tickety:replace-with-a-long-random-password@postgres:5432/tickety

   Use the same password in both database settings. URL-encode reserved characters in DATABASE_URL. Keep .env and credentials out of Git.

3. Start the stack:

   ./deploy.sh docker

4. Create the first administrator:

   docker compose run --rm backend \
     python -m app.backend.bootstrap_admin \
     --name "Project Owner" \
     --email owner@example.com

   Enter a password of at least 12 characters at the hidden prompt. This command works only while the users table is empty; no accounts or sample data are created automatically.

Open localhost:3000 and sign in. Stop the stack with docker compose down; the database is retained.

Configuration

See .env.example for all settings.

• Production: The template defaults to APP_MODE=production, which requires an authenticated session and uses the read-only Freshservice sidecar. Follow the deployment guide for HTTPS, origins, secrets, and infrastructure setup. Keep demo mode local.
• AI: Configure Microsoft Foundry (FOUNDRY_*) or an OpenAI-compatible endpoint (CUSTOM_*), select DEFAULT_MODEL, and set LLM_ALLOWED_PROVIDER_HOSTS for production. Automatic AI workflows and embeddings are off by default.
• SSO: Use the SSO guide to configure Entra ID, Okta, or another OIDC provider.

Development

Backend checks require Python 3.11 and dependencies from requirements.lock. Run in an activated virtual environment:

python -m pip install -r requirements.lock
APP_MODE=demo DATABASE_URL=sqlite:// PYTHONPATH=. \
  python -m unittest discover -s tests -p 'test_*.py' -v

Frontend checks require Node.js 24 or later and npm 12.0.2:

cd app/frontend-next
npm ci
npm test
npm run lint
npm run typecheck
npm run build
npm run verify:production-routes

The production-route check runs against the completed build. Test and browser verification history for the requirements workspace is recorded separately; local checks do not establish that a commit has been deployed to Dev or production.

Documentation

• Deployment — Docker Compose and production Kubernetes
• Development deployment — MicroK8s
• Database migrations
• Requirements gathering — business workflow and user guide
• Requirements runtime — access, capacity, extraction, AI, and audit boundaries
• Requirements verification — regression scenarios and recorded validation limits
• Freshworks app
• RAG operations — indexing, rollout, and rollback

License

MIT