Vantage RuntimeAI · Adopt

Add your own scenario

Start from 15 library check-rides — or generate unlimited custom scenarios from the SQL, DDL, policy, or OpenAPI already in your repo. About 60 seconds, no hand-written YAML.

See the flow ↓Try in SimulatorEditor / MCP

Library vs custom

Library scenarios prove the method quickly. Custom scenarios use the same rubric engine with your fixtures — the id you gate in CI is custom_…, not a library slug.

Library

15 public scenarios

Curated check-rides for common agent workflows — SQL optimization, pipeline triage, support macros, and more. Use as-is or as templates to understand rubric shape.

Ids like de_sql_optimization_v1

Custom

Unlimited from your repo

Point at your tables, routes, refund policy, or API spec. Same deterministic rubrics as library scenarios — no hand-written YAML fixtures.

Ids like custom_abc123…

The 60-second flow

You never author scenario YAML by hand. Attach repo context, generate a draft, save once, then reference the same custom_… id in Simulator, Editor, API, or GitHub Actions.

Which path should I use?
Have Cursor open with a SQL / schema file?
  └─ yes → MCP: runtimeai_generate_scenario  (fastest)

Prefer terminal only?
  └─ yes → runtimeai-ide generate --file …

No install?
  └─ yes → Simulator ?create=1

Wiring into repo / GitHub Actions?
  └─ yes → curl generate + save → Repo → CI guide

Pick your surface

Same generate → save → run pipeline everywhere. Choose the surface that matches where you work today.

Recommended

Cursor / Claude Code (MCP)

Open a SQL or schema file → ask Agent to call runtimeai_generate_scenario. No monorepo clone; no OPENROUTER_API_KEY for generation.

Editor / MCP setup →

Terminal

runtimeai-ide CLI

pip install runtimeai-ide then generate --file … from any project directory.

CLI example ↓

Browser

Simulator create mode

Paste context in the UI — generate, review, and save without installing anything.

Open Simulator →

Automation

HTTP API / curl

POST /api/scenarios/generate then POST /api/scenarios/custom — wire into bootstrap scripts.

curl snippets →

What to paste from your repo

schema

DDL, dbt models, migration files

sql_sample

Broken query + EXPLAIN output

policy

Refund rules, escalation macros, tier entitlements

api_spec

OpenAPI paths, routing rules, error-code matrix

Brief is optional when context blocks are substantial (40+ chars). More specific brief → tighter rubric.

Path 1 — Cursor Agent (MCP)

Install once: pip install runtimeai-ide, wire MCP, restart Cursor. Full setup: Editor / MCP guide.

Agent prompt (open any SQL file): “Use runtimeai_generate_scenario on my open SQL file — draft a partition-pruning check-ride from the attached schema.”

Works without OPENROUTER_API_KEY (hosted generation; server credits apply). Bring a key only when you want to execute and score with runtimeai_run_checkride.

Path 2 — Terminal CLI

pip install runtimeai-ide
runtimeai-ide doctor

runtimeai-ide generate \
  --file models/fact_orders.sql \
  --brief "Agent must fix partition pruning using attached schema" \
  --out runtimeai-draft.json

Omit --brief and the CLI fills one from the file type. Review runtimeai-draft.json, then save via Simulator or API.

Path 3 — Simulator (no install)

Open Simulator create mode — paste schema/SQL/policy, generate, edit the draft, save. Same JSON shape as CLI and API.

Path 4 — HTTP API

curl -sS -X POST https://www.vantageai.cc/api/scenarios/generate \
  -H "Content-Type: application/json" \
  -H "X-Flightdeck-Product: runtimeai" \
  -d @- <<'EOF'
{
  "brief": "Data engineer must fix a partition filter on fact_orders.",
  "product": "runtimeai",
  "contexts": [
    {
      "type": "schema",
      "label": "Orders DDL",
      "content": "CREATE TABLE fact_orders (order_id INT, dt DATE, amount DECIMAL); -- partition column: dt"
    }
  ]
}
EOF

Then POST /api/scenarios/custom with the draft JSON to receive your custom_… id. Copy-paste bootstrap and GitHub Actions snippets on the Repo → CI guide.

For search engines and LLMs
Add RuntimeAI scenarios: 15 library check-rides or generate custom from repo context. Flow: context → generate → save custom_ id → run in Sim/IDE/CI. MCP runtimeai_generate_scenario, runtimeai-ide generate, or POST /api/scenarios/generate.