summaryrefslogtreecommitdiffhomepage
path: root/.rules
diff options
context:
space:
mode:
authorAdam Malczewski <[email protected]>2026-03-24 15:35:23 +0900
committerAdam Malczewski <[email protected]>2026-03-24 15:35:23 +0900
commitcab2ab3f848874bcceb9cadd6257056ba50cf8bb (patch)
tree23c501f624f01b5ea2939156fb214adb389f357c /.rules
parent20f409a21c16c18e614497aebf461282352d03ae (diff)
downloadai-pulse-obsidian-plugin-cab2ab3f848874bcceb9cadd6257056ba50cf8bb.tar.gz
ai-pulse-obsidian-plugin-cab2ab3f848874bcceb9cadd6257056ba50cf8bb.zip
add deletion tool. first tool with user approval prompt.
Diffstat (limited to '.rules')
-rw-r--r--.rules/changelog/2026-03/24/10.md51
1 files changed, 51 insertions, 0 deletions
diff --git a/.rules/changelog/2026-03/24/10.md b/.rules/changelog/2026-03/24/10.md
new file mode 100644
index 0000000..81d22b7
--- /dev/null
+++ b/.rules/changelog/2026-03/24/10.md
@@ -0,0 +1,51 @@
+# Changelog — 2026-03-24 #10
+
+## Mobile "Load failed" Fix (`src/ollama-client.ts`)
+
+- Imported `Platform` from Obsidian for runtime mobile detection.
+- Split `sendChatMessageStreaming` into a dispatcher + mobile/desktop implementations:
+ - Mobile: uses `requestUrl()` (non-streaming) to bypass WebView sandbox.
+ - Desktop: keeps native `fetch()` for real token-by-token streaming.
+- Enhanced error messages with mobile-specific hints.
+- Added `"load failed"` to caught network error patterns in `testConnection`.
+
+## UI: DaisyUI-inspired Collapse (`src/chat-view.ts`, `styles.css`)
+
+- Replaced native `<details>/<summary>` with a checkbox-driven CSS grid collapse.
+- Uses `grid-template-rows: 0fr → 1fr` transition with a rotating arrow indicator.
+- Tightened padding and margins around the collapse for compact layout.
+
+## UI: FAB / Speed Dial (`src/chat-view.ts`, `styles.css`)
+
+- Replaced inline settings/tools buttons with a FAB in the top-right of the messages area.
+- Main trigger: gear icon, rotates 90° on open.
+- Three actions fan downward with staggered animations:
+ - **AI Settings** (sliders icon) — opens the settings modal.
+ - **Tools** (wrench icon) — opens the tools modal.
+ - **Clear Chat** (trash icon) — clears message history and UI.
+- Removed old inline button styles and tools-active coloring.
+
+## UI: Text Selection (`styles.css`)
+
+- Enabled `user-select: text` on messages and tool call bubbles.
+
+## UI: Settings Modal Rename (`src/settings-modal.ts`)
+
+- Changed modal title to "AI Settings".
+
+## Generation Parameters (`src/settings.ts`, `src/settings-modal.ts`, `src/ollama-client.ts`, `src/chat-view.ts`)
+
+- Added `temperature` (default 0.7), `numCtx` (default 4096), `numPredict` (default -1) to settings.
+- Added `ModelOptions` interface and `options` field to `StreamingChatOptions`.
+- Options are now passed to Ollama in the request body for all chat paths.
+- Settings modal shows:
+ - **Temperature** — slider 0–2 with live value display.
+ - **Context Window** — number input with model max shown below.
+ - **Max Output Tokens** — number input (-1 = unlimited).
+- Added `showModel()` function querying `/api/show` to extract the model's context length.
+- Model max label turns red when context window exceeds model limit.
+- Clicking the model max label sets context window to the model's max.
+
+## Disabled GitHub Action (`.github/workflows/lint.yml`)
+
+- Commented out push/PR triggers; added `workflow_dispatch` for manual runs.