[styayur]

Summary

Prepares CodeForge Engine v1.1.0 as a compatibility-focused safety, precision and evidence release.

Baseline safety

• Removes only Resolved entries; Stale entries are retained and reported for review.
• Requires --complete-scan --apply for destructive pruning.
• Blocks changed-file scopes, language filters, submitted/selected mismatches, unavailable relevant lint engines and ambiguous matches.
• Introduces an explicit rule-semantics version independent from the package version.
• Removes the legacy lifecycle-only pruning path.

Rule precision

• Python None comparison fixes use AST operands and avoid string literals/chained comparisons.
• Java empty-catch detection uses the catch body AST and skips documented or nonempty handlers.

Evidence and benchmarks

• Review coverage metadata remains additive and is shown in the CLI JSON and desktop review view.
• Timing summaries add P95.
• Benchmark snapshots record warm-up/sample counts and environment data including Git revision, worktree state, CodeForge and Rust tool versions.
• Release workflow now explicitly sets the release title from the tag.

Release metadata

• Cargo, Cargo.lock, desktop package and Tauri config are aligned to 1.1.0.
• Added user-facing v1.1.0 release notes and retained v1.0.0 evidence separately.

Compatibility and migration

• Baseline schema remains 1; SARIF remains 2.1.0.
• baseline prune is intentionally stricter: scripts must add --complete-scan --apply.
• Old ReviewReport payloads without coverage metadata remain readable.
• MSRV remains Rust 1.90.

Verification

• Rust fmt, Clippy, workspace tests and release CLI build passed locally.
• Frontend lint, typecheck and build passed.
• Baseline dry-run/apply, incomplete-scan refusal and stale-retention tests passed.
• Rule positive/negative tests passed.
• Benchmark P95/environment smoke test passed.
• Changed-file self-review reported zero diagnostics.