feat(plans): update status of coordinated subagents and better harness plans to done

Author: iptoux

.agents/plans/PLANS.md

@@ -13,5 +13,5 @@ Persistent plans for multi-step work on **blxcode**. Individual plans live as Ma
 | done | [terminal-agent-context-handoff.md](terminal-agent-context-handoff.md) | BLXCode Agent kann aktuellen Kontext gezielt per Toolcall an Terminal-Agent-Sessions uebergeben; Bilder werden als Metadaten plus lokale Pfade exportiert |
 | done | [plan-manager.md](plan-manager.md) | Plan Manager fuer `.agents/plans`: Manage-Tab wie Memory Files, plan-linked Tasks gruppiert, Agent-Toolcalls, System-Prompt-Update und shared Context fuer BLXCode Agent plus Terminal-Handoff |
 | planned | [kanban-board-view.md](kanban-board-view.md) | Kanban-View im Plans-Panel fuer alle Plan-Tasks eines Workspaces: Status-Spalten, Drag-and-Drop fuer Karten und Spalten, Spalten ein-/ausblenden und Markdown-Writeback |
-| planned | [coordinated-subagents.md](coordinated-subagents.md) | Coordinated Subagents fuer BLXCode Agent mit Rollen, i18n Live-Subcards, Provider-Reuse, Environment Detection, Shell/Git/Web Toolsets und scoped Toolgruppen |
-| in-progress | [better-harness.md](better-harness.md) | BetterHarness: Shrink system prompt by extracting tool docs into 6 embedded Core Skills; Skills tab gets Core/User sub-tabs |
+| done | [coordinated-subagents.md](coordinated-subagents.md) | Coordinated Subagents fuer BLXCode Agent mit Rollen, i18n Live-Subcards, Provider-Reuse, Environment Detection, Shell/Git/Web Toolsets und scoped Toolgruppen |
+| done | [better-harness.md](better-harness.md) | BetterHarness: Shrink system prompt by extracting tool docs into 6 embedded Core Skills; Skills tab gets Core/User sub-tabs |

CHANGELOG.md

@@ -11,10 +11,28 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 - **Interactive setup scripts**: platform-specific post-clone setup helpers under `scripts/setup/` for Linux, macOS and Windows. They install/check Rust, `wasm32-unknown-unknown`, Trunk, Tauri CLI, Node/npm frontend assets, OS-native Tauri dependencies, and optional verification/bundle builds.
 - **Native release script counterparts**: PowerShell and Command Prompt release entrypoints (`scripts/release.ps1`, `scripts/release.cmd`, `scripts/release-windows.*`) plus a macOS launcher (`scripts/release-macos.sh`). The PowerShell modules under `scripts/release/*.ps1` mirror the Bash release pipeline for version bumps, CHANGELOG finalization, Tauri builds, artifact discovery, and GitHub uploads.
+- **BetterHarness — slim system prompt + Core Skills**: extracted all per-tool documentation from the system prompt into 6 embedded Core Skill files (`file-access`, `memory`, `plans`, `tasks`, `rules-skills`, `harness`). System prompt shrunk from ~470 to ~240 lines; tool docs are served lazily via `skills_read`. New `SkillSourceKind::Core` variant; `list_skills` prepends always-present core skills, `read_skill` serves embedded content, `remove_skill` rejects core names. Skills tab gets Core / User sub-tabs (Install hidden in Core view).
+- **Kanban board view for Plans**: per-workspace board across all plan tasks with status columns, drag-and-drop for cards and columns, show/hide columns, Markdown writeback. New `manage-plans` skill documentation.
+- **Skills tab — expandable SkillCard**: SkillCard now renders as an expandable card with a shadcn-style switch, lazy-loads the skill's `SKILL.md` body on first expand, and shows source-kind icon badges. SkillsTab uses a modern pill switcher for Core / User sub-tabs with counts and improved empty / error states.
+- **Rules tab — expandable RuleCard with inline editing**: RuleCard mirrors the SkillCard layout, can be expanded inline, supports editing rule body, and ships a new rule-creation flow with input validation and UI polish.
+- **Voice recorder**: more reliable input-device name retrieval (resolves cases where the device label fell back to a generic identifier).
+- **CI — PR cargo check workflow**: new GitHub Actions workflow runs `cargo check` for both crates (`blxcode` native Linux, `blxcode-ui` `wasm32-unknown-unknown`) on every pull request, using `dtolnay/rust-toolchain` and `actions/cache`.
+- **Coordinated subagents**: Coordinator can spawn up to 5 parallel subagents (`scout`, `review`, `security_analyst`) via a new `subagents.run` tool that reuses the host agent's provider, API key, model and thinking level. Each subagent runs an isolated message loop with a filtered toolkit, a forced `submit_result` JSON-schema tool, per-run caps (8 iterations, ~20k output tokens) and findings/artifact truncation (max 20/10). Recursion is blocked and `shell_write` is Coordinator-only.
+- **Unified tool dispatch**: new `agent/tool_dispatch.rs` routes Anthropic, OpenRouter and OpenAI-compatible providers through a single dispatcher so all toolgroups and `subagents.run` are wired in one place.
+- **Toolgroup filtering**: `ToolGroup` enum + `registry_filtered` / `render_for_openai_filtered` / `render_for_anthropic_filtered` build per-role catalogs (`environment_read`, `workspace_read`, `diff_read`, `git_read`, `git_write`, `shell_read`, `shell_write`, `web_read`, memory/plans/tasks/rules-skills read/write).
+- **Environment detection**: `environment_detect` tool with session-scoped cache; `shell_exec` and `git_*` reject calls before detection has run, cache is invalidated on workspace switch.
+- **Shell tool**: non-interactive `shell_exec` with workspace CWD, output/time limits, a read-only command allowlist (incl. git subcommand parsing), child-process registry and SIGTERM→SIGKILL (2s) / Windows `TerminateProcess` cancellation. `shell_write` opt-in for Coordinator only.
+- **Git tools**: `workspace_git_status`, `workspace_diff`, plus generic `git_status`, `git_diff`, `git_log`, `git_show`, `git_branch_info`, `git_ls_files` (read) and `git_apply_patch`, `git_add`, `git_commit` (gated behind `git_write`).
+- **Workspace search**: `workspace_search` for ripgrep-style content lookup within the workspace root.
+- **Web tools**: optional `web_search` / `web_fetch` with pluggable backend (Tavily or Brave). API keys live in the OS keyring (`BLXCode` service, accounts `agent:web:tavily` / `agent:web:brave`) with `BLX_TAVILY_API_KEY` / `BLX_BRAVE_API_KEY` env fallback; tools are dropped from the catalog and the `web` skill is flagged `disabled_no_key` when no key is configured. New Settings → Agent → Web Tools block with backend picker and masked key inputs.
+- **Subagent protocol & UI**: new `AgentEvent::Subagent{Started,Step,ToolCall,Finished}` events, `TimelineItem::SubagentGroup` (tolerant serde for older `sessions.json`), inline subcards with live steps and compact tool-call icons, ~50 ms debounce for parallel event bursts.
+- **Core skills**: embedded harness skills `subagents.md`, `environment.md`, `shell.md`, `git.md`, `web.md` plus updated `file-access.md` / `harness.md` (harness-vs-shell boundary). `system_prompt.rs` skill/tool index expanded with subagent activation triggers.
+- **i18n**: subagent UI states, role display names (Scout / Review / Security Analyst with auto-suffix on conflict), all new tool/environment/shell/git/web labels and the web-disabled hint added to every locale.
 
 ### Changed
 
 - **Docs**: README, developer setup, build guide, and `.env.release.example` now document the setup and cross-platform release automation paths.
+- **Dependencies**: upgraded `leptos` 0.7 → 0.8, `leptos_icons` 0.5 → 0.7, `icondata` 0.5 → 0.7. Leptos 0.8 is backward-compatible for the signal / effect / callback / event-listener APIs in use; only Lucide icon renames required code changes (`LuFileEdit → LuFilePenLine`, `LuPlusCircle → LuCirclePlus`, `LuSendHorizonal → LuSendHorizontal`, `LuAlertTriangle → LuTriangleAlert`, `LuTerminalSquare → LuSquareTerminal`, `LuPlayCircle / LuAlertCircle / LuCheckCircle / LuMinusCircle → LuCircle*` variants).
 
 ### Fixed