Use a local target directory

The simplest deployment is a local target/ directory from a dbt project. dbt writes manifest.json and run_results.json there by default after any dbt run, dbt test, or dbt build invocation.

Minimal example

cd my-dbt-project
dbt run
npx @dbt-tools/cli status --dbt-target ./target --json

Configuration

Method	Example
Flag	--dbt-target ./target
Environment variable	DBT_TOOLS_DBT_TARGET=./target

The path can be relative (resolved from the current working directory) or absolute.

Non-default target directories

dbt allows overriding the target directory via --target-path or the target-path key in dbt_project.yml. If your project uses a non-default path, specify it with --dbt-target:

dbt run --target-path ./my-custom-target
npx @dbt-tools/cli status --dbt-target ./my-custom-target --json

Using the Web UI locally

npx @dbt-tools/web --dbt-target ./target

This starts the Web UI on http://localhost:3000 by default. The port and host are configurable. See Web server CLI for flags.

Using MCP locally

npx @dbt-tools/mcp --dbt-target ./target

The MCP server reads the artifacts at startup and holds them in memory for the session. If you run dbt again and want the MCP server to see the new results, restart the server.

Troubleshooting

Symptom	Likely cause	Fix
readiness: unavailable	Path does not contain manifest.json	Confirm you ran dbt and that ./target/manifest.json exists
readiness: manifest-only	run_results.json not present	Run dbt run, dbt test, or dbt build — not just dbt compile
Stale results	Previous run's artifacts still on disk	Re-run dbt; artifacts are overwritten on each run

Related

• S3 — read from remote object storage
• GCS — read from Google Cloud Storage
• Configuration reference