Update devlog and devjournal with session 18 notes

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

DEVLOG.md

@@ -5552,3 +5552,61 @@ Wired Hal chat to the Agent system so it can create new AUTARCH modules on deman
 
 ---
 
+## Session 18 - 2026-03-02
+
+### Phase 4.25 — Hal Chat Fix (Chat/Agent Dual Mode)
+
+**Problem:** All Hal chat messages were routed through the Agent system (`core/agent.py`), which expects structured `THOUGHT:/ACTION:/PARAMS:` responses. Local GGUF models return plain conversational text, causing `_parse_response()` to fail with `ValueError("No ACTION found")` on every message. The agent retried 20 times, exhausted max steps, and returned `Error: '"info"'`.
+
+**Fix — Dual-mode routing:**
+
+**Files Changed:**
+- `web/routes/chat.py` — Split `/api/chat` into `_handle_direct_chat()` (streams tokens via `llm.chat(stream=True)`) and `_handle_agent_chat()` (existing Agent system). Mode selected by `mode` field in POST body (`'chat'` default, `'agent'` for tools).
+- `web/templates/base.html` — Added toggle switch in Hal panel header (Chat ↔ Agent)
+- `web/static/js/app.js` — Added `halAgentMode` flag, `halModeChanged()`, passes `mode` in fetch body
+- `web/static/css/style.css` — Toggle switch CSS (`.hal-mode-switch`, `.hal-mode-slider`, `.hal-mode-label`)
+
+### Phase 4.26 — Agent Graceful Degradation
+
+**Problem:** In Agent mode, models that can't follow the structured format would loop 20 times and error out.
+
+**Fix:**
+
+**Files Changed:**
+- `core/agent.py` — Added `parse_failures` counter. After 2 consecutive `ValueError` from `_parse_response()`, the agent cleans up the raw response (strips ChatML tokens) and returns it as a `task_complete` result instead of continuing to retry. First failure still gets one retry with format correction prompt.
+
+### Phase 4.27 — Frozen Build LLM Fix
+
+**Problem:** Compiled exe reported `llama-cpp-python not installed: No module named 'llama_cpp'` because `llama_cpp` and `llama_cpp_python` were in the PyInstaller excludes list.
+
+**Fix:**
+
+**Files Changed:**
+- `autarch_public.spec` — Removed `llama_cpp`, `llama_cpp_python`, `anthropic` from `excludes` list
+- `setup_msi.py` — Same removal from excludes
+
+### Phase 4.28 — System Tray Icon
+
+**Problem:** No `.ico` file existed — exe had no icon in Explorer/taskbar, and the tray icon relied on Pillow generating one programmatically at runtime.
+
+**Fix:**
+
+**Files Changed:**
+- `autarch.ico` (NEW) — Multi-resolution .ico (16-256px) created from `icon.svg`
+- `icon.svg` (NEW) — SVG source for the AUTARCH icon (anarchy-A in circle, cyberpunk neon style)
+- `core/tray.py` — Added `_get_icon_path()` to find `.ico` in both source and frozen builds. `create_icon_image()` now loads from `.ico` first, falls back to programmatic generation.
+- `autarch_public.spec` — Added `icon=str(SRC / 'autarch.ico')` for both exe targets, added `.ico` to data files
+- `installer.iss` — Added `SetupIconFile=autarch.ico`, `UninstallDisplayIcon`, `IconFilename` on shortcuts
+
+### v1.5.1 Release
+
+**Release:** https://github.com/DigijEth/autarch/releases/tag/v1.5.1
+
+**Assets:**
+- `AUTARCH_Setup.exe` — Inno Setup installer with icon
+- `AUTARCH_v1.5.1_Portable.zip` (51 MB) — Portable build
+
+**Version bumped** in `installer.iss`, `installer.nsi`, `setup_msi.py`.
+
+---
+

devjournal.md

@@ -2324,3 +2324,35 @@ Full Hash Toolkit added as a sub-page under Analyze (sidebar sub-item like Legen
 - Phase 6 (Docker): NOT STARTED
 - Plan file (quizzical-toasting-mccarthy.md) — Threat Monitor + Hal Module Factory: DONE
 
+---
+
+## Session 18 - 2026-03-02: Chat Fix, Tray Icon, v1.5.1 Release
+
+### Fixes
+- **Hal Chat broken** — All messages went through Agent system, models can't follow structured format → `Error: '"info"'`. Fixed by adding Chat/Agent mode toggle: Chat mode streams tokens directly via `llm.chat(stream=True)`, Agent mode uses tool-using Agent system.
+- **Agent infinite retry** — Models that can't produce `THOUGHT/ACTION/PARAMS` looped 20 times. Fixed: after 2 consecutive parse failures, return the raw response as a direct answer.
+- **LLM missing in exe** — `llama_cpp` was in PyInstaller excludes. Removed from `autarch_public.spec` and `setup_msi.py`.
+- **No exe icon** — Created `autarch.ico` from `icon.svg`, wired into PyInstaller spec, Inno Setup installer, and `core/tray.py`.
+
+### New Features
+- Chat/Agent toggle switch in Hal panel header (CSS slider, defaults to Chat mode)
+- `autarch.ico` multi-resolution icon (16-256px) for exe, tray, installer shortcuts
+- `icon.svg` source artwork (anarchy-A cyberpunk neon)
+
+### Release
+- **v1.5.1** pushed to GitHub: https://github.com/DigijEth/autarch/releases/tag/v1.5.1
+- Assets: `AUTARCH_Setup.exe`, `AUTARCH_v1.5.1_Portable.zip` (51 MB)
+
+### Key files modified
+- `web/routes/chat.py` — Dual-mode: `_handle_direct_chat()` + `_handle_agent_chat()`
+- `core/agent.py` — `parse_failures` counter, graceful fallback after 2 failures
+- `core/tray.py` — `_get_icon_path()`, loads `.ico` with fallback
+- `autarch_public.spec` — Icon paths, removed llama_cpp from excludes
+- `web/templates/base.html` — Hal mode toggle switch
+- `web/static/js/app.js` — `halAgentMode`, `halModeChanged()`, mode in POST body
+- `web/static/css/style.css` — Toggle switch styles
+- `installer.iss` — Icon paths, v1.5.1
+- `autarch.ico`, `icon.svg` (NEW)
+
+**Phase status:** Phases 0–4.28 DONE, Phase 6 (Docker) NOT STARTED
+