Resources / Agentic AI / Getting Started

Getting Started

Use the public demo, Agentic Studio, or a workspace project.

Agentic Studio (configure packs)

  1. Log in → open Studio
  2. Models — bind OpenAI, Anthropic, or a custom OpenAI-compatible endpoint
  3. Skills — write prompt overlays and tool allowlists
  4. Agents — pick a model, attach skills, set role/prompt
  5. Workflows — create a pack (pdlc / sdlc / custom) or clone the system PDLC pack
  6. Open the graph editor — add agent/gate/branch nodes, link edges, save
  7. Click Run to open a workspace bound to that pack

Public demo (no login)

  1. Open Agentic AI Platform
  2. Click a PDLC stage and chat via POST /agent/demo

Workspace (run packs)

  1. Dashboard → create or open a project → workspace
  2. AI Chat tab → choose a workflow pack from the dropdown
  3. Use the stage strip or Advance stage
  4. Link to Studio from the workflow bar to edit packs

Example: build a todo app (seeded PDLC)

Step Stage Prompt example
1 Discover "Discover requirements for a personal todo app"
2 Design "Design the architecture and UX"
3 Develop "Implement the todo app with localStorage"
4 Test "Run the app and fix any issues"
5 Deploy "Publish a preview"
6 Iterate "Add due dates to the backlog for next cycle"

API quick start

# Public demo
curl -X POST http://localhost:8000/agent/demo \
  -H "Content-Type: application/json" \
  -d '{"message":"Discover requirements for a todo app","domain":"pdlc","stage":"discover"}'

# List packs (auth)
curl http://localhost:8000/agent/workflows

# Studio — create a skill (auth cookie/session)
curl -X POST http://localhost:8000/studio/api/skills \
  -H "Content-Type: application/json" \
  -d '{"name":"Review","prompt_md":"Review code for bugs.","tool_allowlist":["list_files","read_file"]}'

See API Reference for Studio and agent endpoints.