openclacky 0.7.6 → 0.7.7

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 001070cf8c2d2587620da0669c4cc014bff55930033beafd038ae86ae7374276
-  data.tar.gz: 6c08cf048ab754c8e3be4841496b19b14cf081bdc62cadb313070226cb4e8679
+  metadata.gz: 5c3815310a11bea9ccad7ab6891b4ad85e248635af6919606e8963096699a1d7
+  data.tar.gz: 17f8059179701394a6951966ec0284ee6321750e2651c4de0ecbf37a4542c7fc
 SHA512:
-  metadata.gz: 1bd44e7efcb16f9d9a15bb49c8d253183697107dcbb25f9ad8edc624d1a357d67c7f54f974cc06ae2140f66c0dec4bd2c2d8b50fae93ad3a8cf2453a0ee5f808
-  data.tar.gz: 4328e4ff99db8731e5ed0a1c298c281261da097d214b600e8d6d1b544073055cc35145d76f53853ea30f6f4827f831cd3444129559ff66e10b4d5a23b30e71e0
+  metadata.gz: b7b32ce62e12f9aa12db59b1f6640e4a6234c34c485d9b679df80654dc8722a369687e1fc0b5165a78a12f6210fcd02959133d6d3753c5c22873d071251b07fb
+  data.tar.gz: 20f4498c976579333b7e261490d287a33cec89b888c5ae1a640205c25f48c70c3863a551e1bef6173804b7670970ead88ed9a67702987da97963c0a981cd1e2a

data/.clackyrules

@@ -49,7 +49,11 @@ It provides chat functionality and autonomous AI agent capabilities with tool us
 - **IMPORTANT**: When testing clacky commands or debugging, always use `bundle exec ruby bin/clacky` instead of the global `clacky` command. The global command loads the system-installed gem version (e.g., `openclacky-0.7.0`), not your local development code
 
 ### Tool Development
-When adding new tools:
+**IMPORTANT**: Do NOT add new Ruby Tool classes without careful consideration. Tools are low-level primitives (file I/O, shell, web search, etc.). Before creating a new Tool, always ask: can this capability be achieved with an existing Tool + a Skill (SKILL.md) instead? If yes, use a Skill.
+
+Only add a new Tool when it requires capabilities that no existing tool can provide (e.g., a new API integration, a new system-level operation).
+
+When a new Tool is truly necessary:
 1. Create class in `lib/clacky/tools/`
 2. Inherit from `Clacky::Tools::Base`
 3. Define required class attributes
@@ -57,6 +61,14 @@ When adding new tools:
 5. Add optional `format_call` and `format_result` methods
 6. Require in `lib/clacky.rb`
 
+### Skill Development
+Skills are the preferred way to add new high-level capabilities. A skill is a Markdown instruction file (SKILL.md) that guides the Agent to accomplish a goal using existing Tools.
+
+- Built-in default skills live in `lib/clacky/default_skills/<skill-name>/SKILL.md` (shipped with the gem)
+- Project-level skills live in `.clacky/skills/<skill-name>/SKILL.md`
+- User-level skills live in `~/.clacky/skills/<skill-name>/SKILL.md`
+- **Prefer Skills over new Tools** whenever the task can be composed from existing primitives (write, shell, read, etc.)
+
 ### Agent Behavior
 - TODO manager is for planning only - must execute tasks after planning
 - Always use tools to create/modify files, don't just return code

data/CHANGELOG.md

@@ -7,6 +7,33 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ## [Unreleased]
 
+## [0.7.7] - 2026-03-04
+
+### Added
+- Web UI server with WebSocket support for real-time agent interaction in the browser (`clacky serve`)
+- Task scheduler with cron-based automation, REST API, and scheduled task execution
+- Settings panel in web UI for viewing and editing AI model configurations (API keys, base URL, provider presets)
+- Image upload support in web UI with attach button for multimodal prompts
+- Create Task button in the task list panel for quick task creation from the web UI
+- `create-task` default skill for guided automated task creation
+
+### Improved
+- Web UI frontend split into modular files (`ws.js`, `sessions.js`, `tasks.js`, `settings.js`) for maintainability
+- Web session agents now run in `auto_approve` mode for unattended execution
+- Session management moved to client-side for faster, round-trip-free navigation
+- User message rendering moved to the UI layer for cleaner architecture
+- No-cache headers for static file serving to ensure fresh asset delivery
+
+### Fixed
+- `DELETE`/`PUT`/`PATCH` HTTP methods now supported via custom WEBrick servlet
+- Task run broadcasts correctly after WebSocket subscription; table button visibility fixed
+- Mutex deadlock in scheduler `stop` method when called from a signal trap context
+- `split` used instead of `shellsplit` for skill arguments to avoid parsing errors
+
+### More
+- Add HTTP server spec and scheduler spec with full test coverage
+- Minor web UI style improvements and reduced mouse dependency
+
 ## [0.7.6] - 2026-03-02
 
 ### Added

data/lib/clacky/agent.rb

@@ -774,7 +774,7 @@ module Clacky
 
     # Format user content with optional images
     # @param text [String] User's text input
-    # @param images [Array<String>] Array of image file paths
+    # @param images [Array<String>] Array of image file paths or data: URLs
     # @return [String|Array] String if no images, Array with text and image_url objects if images present
     private def format_user_content(text, images)
       return text if images.nil? || images.empty?
@@ -782,8 +782,13 @@ module Clacky
       content = []
       content << { type: "text", text: text } unless text.nil? || text.empty?
 
-      images.each do |image_path|
-        image_url = Utils::FileProcessor.image_path_to_data_url(image_path)
+      images.each do |image|
+        # Accept both file paths and pre-encoded data: URLs (e.g. from Web UI)
+        image_url = if image.start_with?("data:")
+                      image
+                    else
+                      Utils::FileProcessor.image_path_to_data_url(image)
+                    end
         content << { type: "image_url", image_url: { url: image_url } }
       end
 

data/lib/clacky/cli.rb

@@ -690,5 +690,41 @@ module Clacky
 
 
     end
+
+    # ── server command ─────────────────────────────────────────────────────────
+    desc "server", "Start the Clacky web UI server"
+    long_desc <<-LONGDESC
+      Start a long-running HTTP + WebSocket server that serves the Clacky web UI.
+
+      Open http://localhost:7070 in your browser to access the multi-session interface.
+      Multiple sessions (e.g. "coding", "copywriting") can run simultaneously.
+
+      Examples:
+        $ clacky server
+        $ clacky server --port 8080
+    LONGDESC
+    option :host, type: :string, default: "127.0.0.1", desc: "Bind host (default: 127.0.0.1)"
+    option :port, type: :numeric, default: 7070, desc: "Listen port (default: 7070)"
+    def server
+      require_relative "server/http_server"
+
+      agent_config = Clacky::AgentConfig.load
+
+      # Factory so each new session gets a fresh Client instance
+      client_factory = lambda do
+        Clacky::Client.new(
+          agent_config.api_key,
+          base_url: agent_config.base_url,
+          anthropic_format: agent_config.anthropic_format?
+        )
+      end
+
+      Clacky::Server::HttpServer.new(
+        host:           options[:host],
+        port:           options[:port],
+        agent_config:   agent_config,
+        client_factory: client_factory
+      ).start
+    end
   end
 end

data/lib/clacky/default_skills/create-task/SKILL.md

@@ -0,0 +1,102 @@
+---
+name: create-task
+description: Create a scheduled automated task. User describes what they want to automate, agent creates the task prompt file and optional cron schedule.
+disable-model-invocation: false
+user-invocable: true
+---
+
+# Create Task Skill
+
+## When to Use
+Invoke this skill when a user wants to:
+- Automate something on a schedule (e.g. "send a daily report", "remind me every Monday", "check prices every hour")
+- Create a task that runs automatically ("set up a scheduled task for me")
+- Use the command `/create-task`
+
+## Task & Schedule Storage Format
+
+### Task file: `~/.clacky/tasks/<name>.md`
+Plain text prompt file. When the task fires, the Agent reads this file as its input prompt and executes it autonomously.
+
+Example content of `~/.clacky/tasks/daily_report.md`:
+```
+Check today's GitHub PR list, summarize the status of each PR, and save a daily report to ~/reports/daily_<date>.md
+```
+
+### Schedule file: `~/.clacky/schedules.yml`
+YAML list. Each entry has:
+```yaml
+- name: daily_report         # unique name, matches task name
+  task: daily_report         # references ~/.clacky/tasks/daily_report.md
+  cron: "0 9 * * 1-5"       # 5-field cron: minute hour day month weekday
+  enabled: true
+```
+
+Cron field order: `minute hour day-of-month month day-of-week`
+- `0 9 * * 1-5`  → 09:00 every weekday
+- `0 9 * * *`    → 09:00 every day
+- `0 */2 * * *`  → every 2 hours
+- `*/30 * * * *` → every 30 minutes
+
+## Process
+
+### Step 1: Understand the user's intent
+Ask clarifying questions if needed:
+- What should the task DO? (what action, what goal)
+- How often / when should it run? (if scheduling is needed)
+- Any specific working directory or context?
+
+### Step 2: Generate a task name
+- Lowercase, alphanumeric + underscores only
+- Short and descriptive, e.g. `daily_standup`, `price_monitor`, `weekly_report`
+
+### Step 3: Write the task prompt file
+Use the `write` tool to create `~/.clacky/tasks/<name>.md`.
+
+The prompt content should be:
+- Clear and self-contained (the agent running it has no prior context)
+- Written as a direct instruction to an AI agent
+- Include any relevant details the user provided (URLs, file paths, output format, etc.)
+
+Example:
+```
+write(
+  path: "~/.clacky/tasks/daily_standup.md",
+  content: "Check today's work progress:\n1. Review recent git commits\n2. List open TODOs\n3. Generate a standup summary and print it to the terminal"
+)
+```
+
+### Step 4: Write the schedule (if user wants it automated)
+Read the existing `~/.clacky/schedules.yml` first (if it exists), then append the new entry and write the whole file back.
+
+Use the `write` tool. Example full file:
+```yaml
+- name: daily_standup
+  task: daily_standup
+  cron: "0 9 * * 1-5"
+  enabled: true
+```
+
+If `schedules.yml` already has entries, preserve them and append the new one.
+
+### Step 5: Confirm to the user
+Reply with a clear summary:
+```
+✅ Task created successfully!
+
+📋 Name: daily_standup
+📄 File: ~/.clacky/tasks/daily_standup.md
+⏰ Schedule: Every weekday at 09:00 (cron: 0 9 * * 1-5)
+
+Task prompt:
+> Check today's work progress...
+
+The task will run automatically at the next scheduled time.
+You can also click ▶ in the sidebar to run it immediately.
+```
+
+## Notes
+- If the user only wants a task without a schedule (run manually), skip Step 4
+- Always expand `~` to the actual home path when writing files
+- Task name must be sanitized: only `[a-z0-9_-]`, no spaces
+- The cron scheduler checks every minute; `clacky server` must be running for auto-execution

data/lib/clacky/json_ui_controller.rb

@@ -28,10 +28,6 @@ module Clacky
 
     # === Output display ===
 
-    def show_user_message(content, images: [])
-      emit("user_message", content: content, images: images)
-    end
-
     def show_assistant_message(content)
       return if content.nil? || content.strip.empty?
 
@@ -117,7 +113,7 @@ module Clacky
 
     # === Progress ===
 
-    def show_progress(message = nil, prefix_newline: true)
+    def show_progress(message = nil, prefix_newline: true, output_buffer: nil)
       @progress_start_time = Time.now
       emit("progress", message: message, status: "start")
     end

data/lib/clacky/plain_ui_controller.rb

@@ -107,7 +107,7 @@ module Clacky
 
     # === Progress (no-ops — no spinner in plain mode) ===
 
-    def show_progress(message = nil, prefix_newline: true); end
+    def show_progress(message = nil, prefix_newline: true, output_buffer: nil); end
     def clear_progress; end
 
     # === State updates (no-ops) ===