v0.5.8

v0.5.8 — CRAP Reduction, Render Decomposition, Coverage Baseline

Refactoring

• Decomposed 10 high-complexity functions (execute_with_all_callbacks CC 107→orchestrator+9 helpers, run_headless CC 51, run_commit CC 40, run_doctor CC 31, run_tasks CC 31, App::main_loop CC 39→3, MiseTool::execute CC 37, OpenAiCompat streaming/generate CC 34/31, slash_route CC 32→6, handle_input_panel_key CC 32)
• Split render.rs (4287 LOC) into render/{mod.rs (3215), messages.rs (560), overlays.rs (539)}
• Portable snapshot tests via width-preserving redact_cwd

Tests

• 60 new tests for 0%-coverage functions: LspTool (10), InjectionDetector (22), main.rs helpers (20), TUI model picker (8)
• 1779 total tests passing across workspace; 61.09% line coverage baseline (cargo tarpaulin)
• AstGrepTool production fix: validates action before ensure_binary

Quality

• All clippy warnings resolved (-D warnings clean)
• CRAP scores: all refactored functions dropped off top-20
• Coverage LCOV report generated at coverage.lcov

Docs

• Updated all version refs v0.5.7 → v0.5.8 across README, CHANGELOG, crate READMEs, SSG pages
• Test count updated to 1779+ across docs

Publishing

• All 5 workspace crates published to crates.io at v0.5.8