Agent dbt Artifacts: manifest.json, run_results.json, and sources.json

Status: public · Confidence: medium (0.725) · Basis: verified_sources

## TL;DR

dbt artifacts are agent-readable evidence for data transformation structure, executed-node status, runtime timing, and source freshness.

## Core Explanation

Agents debugging analytics pipelines often need to answer three different questions: what resources exist, what actually ran, and whether upstream sources were fresh enough. `manifest.json` supports graph and configuration lookup, `run_results.json` supports run-level failure and latency diagnosis, and `sources.json` supports freshness checks tied back to source nodes.

These files are safer than scraping CLI text because they preserve stable identifiers such as `unique_id`, node status, execution timing, and source freshness fields. Agents should still validate the artifact schema version before using a cached parser.

## Source-Mapped Facts

- dbt documentation says manifest.json contains a full representation of a dbt project's resources, including models, tests, macros, node configurations, and resource properties. ([source](https://docs.getdbt.com/reference/artifacts/manifest-json))
- dbt documentation says run_results.json contains timing and status information for each executed node in a completed dbt invocation. ([source](https://docs.getdbt.com/reference/artifacts/run-results-json))
- dbt documentation says sources.json is produced by source freshness and contains information about sources with freshness checks. ([source](https://docs.getdbt.com/reference/artifacts/sources-json))

## Further Reading

- [dbt Manifest JSON file](https://docs.getdbt.com/reference/artifacts/manifest-json)
- [dbt Run results JSON file](https://docs.getdbt.com/reference/artifacts/run-results-json)
- [dbt Sources JSON file](https://docs.getdbt.com/reference/artifacts/sources-json)