Code and artifacts flagged for future cleanup. Each item is actionable — remove, replace, rewrite, or consolidate. Nothing here is currently in progress; the list exists so we can pick up simplification work when timing allows.
File is an empty template. CI populates
legal.benchmark_license.verified_by_original_authors in entries based on this
file, but with no data source it's always false.
Dependent code to remove alongside:
check_verified_by_original_authors()in scripts/quick_check.py (~line 332)verified_by_original_authorsfield in registry-schema.json (~line 87-90)
Action: either (a) populate with a sourcing policy + data, or (b) remove the file, the check function, and the schema field in one PR.
registry-schema.json:191-201 and scripts/quick_check.py:311-327
quick_check.py attempts to introspect action_space from a reset task but the
logic fails silently — output is ::notice::Could not introspect action_space
and no entry in entries/ has it populated.
Action: remove the schema definition and the introspection attempt. If a consumer ever needs it, implement properly with explicit error handling.
- docs/.gitkeep —
docs/has generated content - entries/.gitkeep —
entries/has real YAML files - stress-results/.gitkeep — populated on first slow-check run
Action: delete all three.
README.md:43 and README.md:82 say:
CI will validate your entry and auto-merge if it passes. No human review needed.
But openspec/specs/ci/spec.md:58 and openspec/specs/ci/spec.md:114 say:
No auto-merge — a human approves before anything lands on main.
Action: pick one and update the other. Recommendation: update README to match spec (safer default; easy to relax later).
Points at cube-standard/blob/design/registry-specs/design/registry_specs.md
— a branch that doesn't exist. The content has moved to
cube-registry/openspec/specs/.
Action: replace with a link to openspec/specs/{entry,ci,site}/spec.md or
remove the Specification section entirely.
scripts/quick_check.py, scripts/slow_check.py, scripts/update_owners.py, scripts/health_check.py
Every script creates its own YAML() instance with near-identical settings. Drift
risk as settings diverge silently.
Action: add scripts/_yaml.py with:
from ruamel.yaml import YAML
def _yaml() -> YAML:
y = YAML()
y.preserve_quotes = True
y.indent(mapping=2, sequence=4, offset=2)
return y
def load(path): ...
def dump(data, path): ...Import from every script. Consistent round-trip, single source of truth for YAML config.