| name | dec-bench-quickstart |
| description | Validate your DEC Bench setup and run your first scenario. Use when a user says "get started", "first run", "test my setup", or "run a quick benchmark". |
DEC Bench Quickstart
Validate the user's setup and run their first benchmark scenario.
Step 1: Validate setup
Check that everything is in place before running:
- Docker: run
docker info — if it fails, tell the user to start Docker Desktop
- CLI: run
dec-bench list — if it fails, tell the user to install with curl -fsSL https://decbench.ai/install.sh | sh
- Repo: check that
scenarios/foo-bar-csv-ingest/scenario.json exists — if not, the user isn't in the repo root
- API key: check that
ANTHROPIC_API_KEY is set (or OPENAI_API_KEY / CURSOR_API_KEY depending on which agent they want to use) — if not, ask them to export it
If anything is missing, tell the user exactly what to fix and stop. Don't proceed with a broken setup.
Step 2: Build and run
Once setup is validated, run the first scenario:
dec-bench build --scenario foo-bar-csv-ingest
dec-bench run --scenario foo-bar-csv-ingest
Step 3: Show results
dec-bench results --latest --scenario foo-bar-csv-ingest
Explain the output: which gate the agent reached, the normalized score, and where the artifacts are.
Done
The user has a working setup and their first result. Point them to:
dec-bench list to see all available scenarios
- The
dec-bench-run skill to run more scenarios
- The
dec-bench-create-scenario skill to create their own