docs: update test count and version refs for v0.5.7

CHANGELOG.md

@@ -15,7 +15,7 @@
 - Hindi welcome popup replacing maki splash screen; cyan-accent default theme
 
 ### Tests
-- 1627+ tests passing across 18 suites (up from 988+)
+- 1643+ tests passing across 18 suites (up from 988+)
 - Added SubagentCoreTests, ModelCatalogTests, IrcHubTests, CommitQueuePanel, CoordinatorTests, GoalLoopRuntime test suites
 
 ### Fixes

README.md

@@ -57,7 +57,7 @@ The core claim:
 
 The short version: **pawan is the vibe-coding runtime for people whose production language already fights bad code for them.** If you're using Rust in anger, that's the feature you want.
 
-## What's New in v0.5.6
+## What's New in v0.5.7
 
 ### TUI reliability — /theme, input contrast, status polish
 - `/theme nord`, `/theme onedark`, `/theme gruvbox`, and other argument-bearing slash commands now submit correctly when pressing Enter
@@ -186,7 +186,7 @@ pawan/
 - **Iteration budget awareness** — warns model when 3 tool iterations remain
 - **Think-token stripping** — strips `<think>...</think>` from content and tool arguments
 
-## TUI (v0.5.6)
+## TUI (v0.5.7)
 
 - **Welcome screen** — model, version, workspace on first launch. Press any key to dismiss.
 - **Command palette** (`Ctrl+P`) — fuzzy-searchable slash commands with model presets
@@ -210,7 +210,7 @@ pawan/
 - **Model selector** — interactive model selection with search and filtering
 - **Session browser** — browse, load, and manage saved sessions with fuzzy search
 - **Auto-save** — automatic session saving at configurable intervals
-- **Comprehensive testing** — full workspace tests plus 1627+ tests passing across 18 suites
+- **Comprehensive testing** — full workspace tests plus 1643+ tests passing across 18 suites
 
 ### Intelligence (2026-04-08)
 

crates/pawan-cli/README.md

@@ -1,14 +1,14 @@
-# pawan v0.5.6
+# pawan v0.5.7
 
 Pawan (पवन) — CLI coding agent with pluggable LLM backends, 34 tools, and cross-session memory.
 
-## What's New in v0.5.6
+## What's New in v0.5.7
 
 - **Slash command reliability** — `/theme <name>` and other argument-bearing slash commands submit correctly when pressing Enter
 - **Readable input placeholder** — textarea placeholder and reset paths use active theme colors instead of low-contrast defaults
 - **Status bar polish** — model, tokens, context percentage/bar, iteration, and clock are separated and spaced clearly
 - **Regression coverage** — added key-event and Ratatui TestBackend tests for `/theme`, placeholder styling, and status formatting
-- **Test suite expansion** — 53 new TUI types tests covering format parsing, strip_reasoning_tags, ContentBlock, ToolBlockState; 1627+ total workspace tests across 18 suites
+- **Test suite expansion** — 53 new TUI types tests covering format parsing, strip_reasoning_tags, ContentBlock, ToolBlockState; 1643+ total workspace tests across 18 suites
 
 ## What's New in v0.5.0
 

crates/pawan-core/README.md

@@ -1,11 +1,11 @@
-# pawan-core v0.5.6
+# pawan-core v0.5.7
 
 Core library for the Pawan CLI coding agent. Contains the agent engine, tool system, configuration, and healing/recovery logic.
 
-## What's New in v0.5.6
+## What's New in v0.5.7
 
 - **Release alignment** — workspace crates now publish together at `0.5.7`; the CLI includes the `/theme` Enter fix, readable placeholder styling, status bar spacing, and expanded TUI regression tests
-- **Test suite expansion** — 29 new compaction tests (edge cases, strategies, keywords, summary parsing) and 16 new eruka bridge tests (serde, JSON parsing, disabled paths); 1627+ total workspace tests across 18 suites
+- **Test suite expansion** — 29 new compaction tests (edge cases, strategies, keywords, summary parsing) and 16 new eruka bridge tests (serde, JSON parsing, disabled paths); 1643+ total workspace tests across 18 suites
 
 ## What's New in v0.5.0
 

docs/content/guide.md

@@ -73,15 +73,15 @@ pawan
 pawan explain src/main.rs
 ```
 
-## What's New in v0.5.6
+## What's New in v0.5.7
 
 ### TUI Reliability & Polish
 
 - **Slash command submission** — `/theme <name>` and other argument-bearing slash commands now submit on Enter instead of being intercepted by the inline picker
 - **Readable input placeholder** — placeholder styling is theme-aware at startup, after resets, and after theme switches
 - **Polished bottom status bar** — model, tokens, context percentage/bar, iteration, and timestamp are visually separated
 - **Expanded TUI tests** — key-event regressions and Ratatui TestBackend assertions cover `/theme`, placeholder styling, and status formatting
-- **Test suite expansion** — 98 new tests added (29 compaction, 16 eruka bridge, 53 TUI types); 1627+ total tests passing across 18 suites
+- **Test suite expansion** — 98 new tests added (29 compaction, 16 eruka bridge, 53 TUI types); 1643+ total tests passing across 18 suites
 
 ## Configuration
 

docs/content/triage.md

@@ -109,7 +109,7 @@ Updated 2026-04-29:
 - **Fastest task completion**: Qwen3.5 122B (13.6s for healing module review)
 - **Highest SWE-bench**: MiniMax M2.5 (80.2%)
 - **16 data structures** in grind workspace
-- **1627+ tests** passing across 18 suites; major test additions in v0.5.7 (SubagentCore, ModelCatalog, IrcHub, CommitQueuePanel, Coordinator, GoalLoopRuntime)
+- **1643+ tests** passing across 18 suites; major test additions in v0.5.7 (SubagentCore, ModelCatalog, IrcHub, CommitQueuePanel, Coordinator, GoalLoopRuntime)
 - **34 tools** in 3 tiers (Core/Standard/Extended) with auto-install via mise
 - **Multi-model thinking support**: Qwen (`enable_thinking`), Gemma (`enable_thinking`), GLM (`enable_thinking` + `clear_thinking`), Mistral Small 4 (`reasoning_effort`), DeepSeek (`thinking`)
 - **Token budget tracking**: thinking vs action token split visible in TUI and CLI