Problem
Several Lab WP Codebox runs failed before the agent task could execute because Playground startup could not fetch WordPress release metadata:
Unable to resolve Playground startup asset wordpress-release-metadata for WordPress latest from https://api.wordpress.org/core/version-check/1.7/?channel=beta: fetch failed
This showed up during WPSG Lab pilots while switching provider paths. It is independent of the model/provider choice and turns transient network/metadata fetch failures into full task failures.
Expected
WP Codebox should make this startup path more resilient for Lab cooking:
- cache WordPress release metadata/startup assets between runs when safe,
- retry transient fetch failures with useful backoff,
- optionally allow the runner/plan to pin a known WordPress version or use a local cached release asset,
- surface a concise diagnostic that distinguishes infrastructure fetch failure from agent/model failure.
Evidence From Today
Observed in WPSG pilot runs including:
site-generation-loop-20260615c
site-generation-loop-20260615e
Representative diagnostic:
agent_runtime.workload: Unable to resolve Playground startup asset wordpress-release-metadata for WordPress latest from https://api.wordpress.org/core/version-check/1.7/?channel=beta: fetch failed
Artifacts referenced by Homeboy included WP Codebox artifact directories such as:
/home/chubes/Developer/.tmp/homeboy-wp-codebox-artifacts-C7Pk4Q
/home/chubes/Developer/.tmp/homeboy-wp-codebox-artifacts-tzyP4h
Why This Matters
For Lab cooking, startup metadata fetch should not be a frequent blocker. It also obscures whether the actual provider/model path is working because the failure happens before useful agent execution evidence exists.
Acceptance Criteria
- Transient
api.wordpress.org metadata fetch failures are retried or served from a safe cache.
- The final diagnostic includes whether the failure was cache miss, network fetch failure, or invalid metadata.
- A plan/runtime option exists to pin or reuse a cached WordPress release for repeatable Lab runs.
Problem
Several Lab WP Codebox runs failed before the agent task could execute because Playground startup could not fetch WordPress release metadata:
Unable to resolve Playground startup asset wordpress-release-metadata for WordPress latest from https://api.wordpress.org/core/version-check/1.7/?channel=beta: fetch failedThis showed up during WPSG Lab pilots while switching provider paths. It is independent of the model/provider choice and turns transient network/metadata fetch failures into full task failures.
Expected
WP Codebox should make this startup path more resilient for Lab cooking:
Evidence From Today
Observed in WPSG pilot runs including:
site-generation-loop-20260615csite-generation-loop-20260615eRepresentative diagnostic:
agent_runtime.workload: Unable to resolve Playground startup asset wordpress-release-metadata for WordPress latest from https://api.wordpress.org/core/version-check/1.7/?channel=beta: fetch failedArtifacts referenced by Homeboy included WP Codebox artifact directories such as:
/home/chubes/Developer/.tmp/homeboy-wp-codebox-artifacts-C7Pk4Q/home/chubes/Developer/.tmp/homeboy-wp-codebox-artifacts-tzyP4hWhy This Matters
For Lab cooking, startup metadata fetch should not be a frequent blocker. It also obscures whether the actual provider/model path is working because the failure happens before useful agent execution evidence exists.
Acceptance Criteria
api.wordpress.orgmetadata fetch failures are retried or served from a safe cache.