🔧🧑‍💻 - Optimize manual RepoWise analysis and complete bounded dead code and history scans #644

Is there an existing issue for this?

• I have searched the existing issues.

What area does this task affect?

CI/CD

Task description

Optimize the manual RepoWise workflow introduced by #639 so that dead-code and security analysis complete useful, explicitly scoped reports within deliberate time and memory budgets.

Profile index creation, health reporting, dead-code analysis, and full-history security separately. Remove redundant computation where the pinned tool supports it, reuse valid indexed results, and process historical data in bounded units rather than collecting the complete history in memory. Choose the implementation from measurements and the pinned tool's actual interfaces.

Acceptance criteria:

• RepoWise remains manual-only and independent of required CI/merge-queue checks.
• Health Markdown/JSON, dead-code JSON, security JSON, immutable revision/scope evidence, timings, and the Job Summary are published on a successful final-revision run.
• Dead-code and security finish within explicit bounded budgets on this repository. Record before/after timings and memory evidence; removing all limits or marking an incomplete scan complete is not a solution.
• Preserve full reachable-history security coverage, including secrets introduced and later removed, without loading all historical blob contents into memory. Any intentional exclusions or supported scope choices are explicit in reports and documentation.
• Reuse deterministic index/analysis results where valid and avoid repeated expensive work when producing multiple report formats.
• Preserve per-run isolation, cleanup, least-privilege permissions, safe subprocess execution, and truthful failure/partial status.
• Add focused workflow/orchestration regressions and fixtures covering bounded history processing, complete coverage, malformed output, failures/timeouts, and report publication. Update affected SDD/catalog evidence and operator documentation.
• Verify applicable validators and the native manual workflow on the final revision; link actual complete scan reports and measured results.

Current issues or inefficiencies

Manual run 39573977241 on commit 3ede9d17ec5a92b9196f0d947f91d17b29e13ecc completed its job in 20m 37s. Index creation alone reported 8m 6s, and health analysis ran separately for Markdown and JSON.

The retained artifact contains valid health reports, but dead-code hit its 90-second timeout (exit 124), and full-history security failed with MemoryError under the 2 GiB virtual-memory limit (exit 1). Both correctly remain incomplete.

The user requests optimizing the manual workflow and resolving these resource-driven incomplete scans, while retaining useful analysis rather than hiding failures.

Expected impact

Shorter on-demand analysis, complete dead-code/security evidence, and bounded resource use without putting RepoWise back on the automatic CI path.

Alternatives considered

Simply increasing every timeout/memory cap keeps duplicated work and unbounded historical materialization. Narrowing history silently would lose audit coverage. Prefer measured reuse and bounded processing; adjust budgets only when supported by measured resource requirements.

Additional context or comments

• Follow-up to 🔧 - Run RepoWise only on demand to avoid CI bottlenecks #639 and PR #641.
• Workflow: .github/workflows/repowise.yml.
• Workflow contracts: scripts/validate-workflow-contract.cjs and src/tooling/__tests__/validate_workflow_contract.test.ts.
• Current tool pin: RepoWise 0.45.0.
• Relevant SDD: specs/merge-queue-readiness.md; update its manual-only contract and affected catalog evidence.