P
Prairie sand Dinosaur
The current Harness Artifact Repository (HAR) metadata API (PUT /har/api/v2/metadata) fully replaces the entire metadata array on every update. This behavior prevents incremental updates, makes partial metadata changes impossible, and introduces unnecessary complexity for teams that need additive or in-place metadata updates.
A new capability is requested to allow metadata to be updated, merged, or patched, rather than overwritten.
Problem Description
This “replace-only” model has several limitations:
  1. No ability to set metadata incrementally
Tools cannot safely add properties without first fetching and re-sending the entire metadata list.
  1. No atomicity
Concurrent updates from separate automation tools risk overwriting each other’s metadata.
  1. Not aligned with common artifact property semantics
Systems like JFrog Artifactory, Nexus, AWS ECR, and others support independent key-value updates — additive, mutable properties.
  1. Increased complexity for customers
To simulate incremental updates, customers must:
• GET existing metadata
• Merge keys manually
• PUT full metadata back
• Handle race conditions and conflicting updates
This logic is duplicated across many pipelines and teams.