# Convalesce

> Convalesce self-heals your data infrastructure.

Convalesce self-heals your data infrastructure. Agents pick up a failed run, trace its blast radius, and return a fix with the evidence behind it.

## Facts
- Website: https://convalesce.io
- Categories: Developer Tools
- LotsDirectory page: https://lots.directory/products/convalesce
- Listing state: Unclaimed. We listed this product from its public website, and its owner has not claimed it yet.

## Detailed info

## What is Convalesce?

Convalesce is an automated incident investigation and self-healing platform for data infrastructure that traces failures and drafts pull requests with code fixes.

## What Convalesce does

When an orchestrator run fails, Convalesce captures the failed run, including exceptions, task state, and concurrent runs. It gathers context from connected tools such as table shapes, lineage, and repository code, working out the cause and opening a pull request with the fix and evidence for an engineer to review.

## Who Convalesce is for

Data teams and engineers running data stacks with orchestrators, warehouses, models, and dashboards who want automated incident investigations.

## How Convalesce works

Convalesce connects to orchestrators like Airflow and warehouses like Snowflake. When a job fails, automated agents collect execution context and lineage, trace downstream impacts, and open a pull request in GitHub for an engineer to review and merge.

## Key features

- Captures failed orchestrator runs including task states, exceptions, and retries.
- Maps data lineage and schema shapes across tables, models, and dashboards.
- Inspects commits, queries, and models to determine what code ran and what changed in repository commits.
- Opens pull requests with verified code fixes and supporting evidence.
- Executes capped, masked read-only queries without storing underlying data rows.
- Integrates with Airflow, Dagster, Prefect, Snowflake, dbt, Spark, Postgres, and GitHub.

## Facts from convalesce.io, checked 15 October 2026

- Investigates data pipeline failures and opens pull requests with fixes and evidence.
- Reads schema definitions, table lineage, and run metadata without storing rows.
- Connects with Airflow, Snowflake, dbt, and GitHub.

Written from Convalesce's own website ([Homepage](https://convalesce.io/), [Features](https://convalesce.io/use-cases), [About](https://convalesce.io/about)). The maker can claim this listing to edit it.

## FAQ

### Does Convalesce apply code fixes automatically without human review?
No. Convalesce opens a pull request with attached evidence against the team repository for engineers to review and merge.

### What data access does Convalesce require?
Convalesce requires read access to run metadata, schemas, row counts, and lineage, running capped read-only queries when needed.

### Which data tools and orchestrators does Convalesce support?
Convalesce integrates with tools including Airflow, Dagster, Prefect, dbt, Snowflake, Databricks, Postgres, Tableau, and GitHub.

### Does underlying warehouse data leave the team environment?
Only the specific incident context bundle needed to investigate the failed run is processed, without copying warehouse data.

---
Source: https://lots.directory/products/convalesce · Machine-readable context by LotsDirectory.