Opcore 0.2.0

Opcore 0.2.0 makes the default robustness loop lighter on large repositories and strengthens validation correctness. This release reduces several major sources of unnecessary in-memory data while keeping coverage and degraded checks explicit.

Highlights

• Bounds product scan validation output to summaries, counts, failed check IDs, and diagnostic samples instead of retaining and serializing an unrestricted result payload.
• Makes validation file views lazy so path-specific checks do not enumerate every visible file unless they need the complete file universe.
• Sends committed clone-analysis inputs as path references and reserves content-bearing overlays for changed or hypothetical files.
• Replaces the Python syntax heuristic with parser-backed ast.parse validation, including structured locations and an honest unsupported result when no Python interpreter is available.
• Adds native validation-policy parity and tightens the package guardrails around it.

Memory scope

These changes materially reduce memory pressure in scan and validation paths, but they do not complete the broader large-monorepo memory program. Graph build/query working sets, inspect/edit language-service isolation, bounded history and sidecar payloads, and regression harness work remain tracked separately.

Upgrade

npm install --global opcore@0.2.0

Or run it without a global install:

npx opcore@0.2.0 --version

Full comparison: v0.1.0...v0.2.0