v0.1.1

.gitignore

@@ -183,9 +183,9 @@ cython_debug/
 .abstra/
 
 # Visual Studio Code
-#  Visual Studio Code specific template is maintained in a separate VisualStudioCode.gitignore 
+#  Visual Studio Code specific template is maintained in a separate VisualStudioCode.gitignore
 #  that can be found at https://github.com/github/gitignore/blob/main/Global/VisualStudioCode.gitignore
-#  and can be added to the global gitignore or merged into this file. However, if you prefer, 
+#  and can be added to the global gitignore or merged into this file. However, if you prefer,
 #  you could uncomment the following to ignore the entire vscode folder
 # .vscode/
 
@@ -224,3 +224,4 @@ bmad-custom-modules-src/
 bmad-custom-src/
 .quint/
 .DS_Store
+.scripts/

.gobby/tasks_meta.json

@@ -1,4 +1,4 @@
 {
-  "content_hash": "bf7be0a664be1e1dcc114277ac3b6fcdf21f842686240c1a1e9764e9604bd9b5",
-  "last_exported": "2026-01-07T21:32:34.730077+00:00"
+  "content_hash": "2fdb70d5a95587d42346a85e4cda7ad012db61cb0f7acd4279355c62f4b4db5e",
+  "last_exported": "2026-01-08T17:28:05.141009+00:00"
 }

AGENTS.md

@@ -224,9 +224,9 @@ Statuses: `active` → `stale` → `merged` → `abandoned`
 Step-based workflows enforce tool restrictions:
 
 ```bash
-uv run gobby workflow list       # Available workflows
-uv run gobby workflow set NAME   # Activate workflow
-uv run gobby workflow status     # Current state
+uv run gobby workflows list       # Available workflows
+uv run gobby workflows set NAME   # Activate workflow
+uv run gobby workflows status     # Current state
 ```
 
 Built-in: `plan-execute`, `test-driven`, `plan-act-reflect`

CLAUDE.md

@@ -152,9 +152,9 @@ Statuses: `active` → `stale` → `merged` → `abandoned`
 Step-based workflows enforce tool restrictions:
 
 ```bash
-uv run gobby workflow list       # Available workflows
-uv run gobby workflow set NAME   # Activate workflow
-uv run gobby workflow status     # Current state
+uv run gobby workflows list       # Available workflows
+uv run gobby workflows set NAME   # Activate workflow
+uv run gobby workflows status     # Current state
 ```
 
 Built-in: `plan-execute`, `test-driven`, `plan-act-reflect`

GEMINI.md

@@ -0,0 +1,104 @@
+# Gobby - Project Context & Instructions
+
+## Project Overview
+
+**Gobby** is a local-first daemon that unifies AI coding assistants (Claude Code, Gemini CLI, Codex) into a persistent, orchestrated environment. It provides long-term memory, session management, and an MCP (Model Context Protocol) proxy with lazy tool discovery.
+
+* **Core Tech:** Python 3.11+, FastAPI, FastMCP, SQLite, Click.
+* **Key Concept:** "Unified Agent Manager" - Gobby sits between the AI CLI and the OS/Tools.
+
+## Environment & Setup
+
+This project uses **[uv](https://github.com/astral-sh/uv)** for dependency management.
+
+### Installation
+
+```bash
+uv sync
+```
+
+### Running the Daemon
+
+```bash
+# Start daemon (verbose for dev)
+uv run gobby start --verbose
+
+# Check status
+uv run gobby status
+```
+
+## Development Workflow
+
+### Quality Checks (Mandatory)
+
+All changes must pass these checks.
+
+```bash
+# Linting & Formatting
+uv run ruff check src/
+uv run ruff format src/
+
+# Type Checking (Strict)
+uv run mypy src/
+
+# Testing
+uv run pytest
+```
+
+### Directory Structure
+
+* `src/gobby/cli/`: Click CLI entry points.
+* `src/gobby/runner.py`: Main daemon process runner.
+* `src/gobby/servers/`: HTTP (:8765) and WebSocket (:8766) servers.
+* `src/gobby/hooks/`: Central hook management logic.
+* `src/gobby/mcp_proxy/`: Logic for connecting to downstream MCP servers.
+* `src/gobby/storage/`: SQLite database layer (`~/.gobby/gobby.db`).
+
+## Architecture Quick Reference
+
+1. **CLI Hook** (from Claude/Gemini) -> **Hook Script** -> **HTTP POST** (`/api/v1/hooks/...`)
+2. **Daemon** (`HookManager`) processes event -> Updates **Session** / **Memory**.
+3. **MCP Proxy**:
+   * Tools are *not* loaded at startup.
+   * `list_tools` fetches metadata only.
+   * `get_tool_schema` fetches full schema on-demand.
+
+## Agent Protocol (CRITICAL)
+
+> "If it's not a task, it didn't happen."
+
+You are operating within a Gobby-enabled environment. You **must** use the `gobby-tasks` system to track your work. Do not rely on chat history or loose files.
+
+### 1. Start of Session
+
+1. **Check Context:**
+   * `mcp_call_tool("gobby-tasks", "list_ready_tasks", {})`
+   * `mcp_call_tool("gobby-tasks", "get_task", {"task_id": "..."})` (if ID is known)
+
+2. **Define Work:**
+   * If new request: `mcp_call_tool("gobby-tasks", "create_task", {"title": "..."})`
+   * If complex: Break down into subtasks using `parent_task_id`.
+
+3. **Link Session:**
+   * `mcp_call_tool("gobby-tasks", "link_task_to_session", {})`
+
+### 2. Execution Loop
+
+* **Update Status:** Mark task as `in_progress`.
+* **Dependencies:** If blocked, use `add_dependency`.
+* **Bugs:** Found a side-issue? `create_task` (don't get distracted).
+
+### 3. End of Session ("Landing the Plane")
+
+* **Close Tasks:** `mcp_call_tool("gobby-tasks", "close_task", {"task_id": "...", "reason": "completed"})`
+* **Clean Up:** Don't leave tasks `in_progress` if you stopped working on them.
+
+## MCP Tool Usage Guide
+
+Gobby uses a proxy pattern for tools.
+
+* **List Tools:** `mcp_list_tools(server_name="gobby-tasks")`
+* **Get Schema:** `mcp_get_tool_schema(server_name="gobby-tasks", tool_name="create_task")`
+* **Call Tool:** `mcp_call_tool(server_name="gobby-tasks", tool_name="create_task", arguments={...})`
+
+*Note: Replace "gobby-tasks" with "gobby-memory" or "gobby-skills" for other internal domains.*