Group the Goose baseline and serialize it as YAML #682

Closed
opened 2026-07-24 01:09:29 +00:00 by coilyco-ops · 3 comments
Member

Follow-up to #656. Replace the flat JSON before snapshot with deterministic YAML grouped first by eager or lazy delivery, then by component kind. Preserve component evidence, payload hashing, readable summaries, and before-versus-after comparison. Keep legacy JSON snapshots readable when practical. Update fixtures and the walkthrough, then land the new before artifact.

Follow-up to #656. Replace the flat JSON before snapshot with deterministic YAML grouped first by eager or lazy delivery, then by component kind. Preserve component evidence, payload hashing, readable summaries, and before-versus-after comparison. Keep legacy JSON snapshots readable when practical. Update fixtures and the walkthrough, then land the new before artifact.
Author
Member

Refinement: retain per-skill visibility while collapsing each skill to one record. Nest eager frontmatter and lazy body/resource measurements under that record so the YAML preserves load-stage boundaries without repeating skill metadata.

Refinement: retain per-skill visibility while collapsing each skill to one record. Nest eager frontmatter and lazy body/resource measurements under that record so the YAML preserves load-stage boundaries without repeating skill metadata.
Author
Member

Final readability refinement: each skill keeps only class, eager tokens, lazy tokens, resource count, and one combined payload hash. Kind and stage totals remain once at the top. Per-skill bytes, paths, component counts, body/resource splits, and separate hashes are redundant for the comparison conclusions.

Final readability refinement: each skill keeps only class, eager tokens, lazy tokens, resource count, and one combined payload hash. Kind and stage totals remain once at the top. Per-skill bytes, paths, component counts, body/resource splits, and separate hashes are redundant for the comparison conclusions.
Author
Member

The real render showed the per-skill hash was itself the remaining visual noise. The final schema keeps four per-skill conclusions only: class, eager tokens, lazy tokens, and resource count. The top-level payload hash now fingerprints the full underlying component hashes, so content sensitivity remains without repeating hashes in every row.

The real render showed the per-skill hash was itself the remaining visual noise. The final schema keeps four per-skill conclusions only: class, eager tokens, lazy tokens, and resource count. The top-level payload hash now fingerprints the full underlying component hashes, so content sensitivity remains without repeating hashes in every row.
Sign in to join this conversation.
No milestone
No project
No assignees
1 participant
Notifications
Due date
The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set

Reference
coilyco-flight-deck/agentic-os#682
No description provided.