diff --git a/skills/swarm-orchestration/SKILL.md b/skills/swarm-orchestration/SKILL.md
new file mode 100644
index 0000000..7234c67
--- /dev/null
+++ b/skills/swarm-orchestration/SKILL.md
@@ -0,0 +1,90 @@
+---
+name: Swarm Orchestration
+description: Pattern for 1-to-many control of agent instances using tmux and session resumption.
+---
+
+# 1-to-Many Agent Swarm Orchestration Skill
+
+## Overview
+This skill defines a pattern for an orchestrator agent to spawn, track, and control multiple sub-agent instances (a "swarm") using `tmux` for environment isolation and command-line interfaces supporting session resumption.
+
+## Prerequisites
+- `tmux` installed on the host system.
+- An interactive CLI or agent tool that supports:
+    - Running in a terminal.
+    - Exposing a unique Session/Conversation ID.
+    - Resuming sessions via targeting flags (e.g., `-conversation="<ID>"`).
+
+## Procedure
+
+### 1. Initialize Swarm Environment
+Create a `tmux` session with multiple windows or panes, depending on the desired layout.
+
+**Windows Layout (Clean separation):**
+```bash
+tmux new-session -d -s <session_name> -n agent1
+tmux new-window -t <session_name> -n agent2
+...
+```
+
+**Grid Layout (All-in-one view):**
+```bash
+tmux new-window -t <session_name> -n grid
+tmux split-window -t <session_name>:grid
+tmux select-layout -t <session_name>:grid tiled
+```
+
+### 2. Spawn Agents
+Send execution commands to the specific panes/windows.
+
+```bash
+tmux send-keys -t <session_name>:<target> "/path/to/cli --args" C-m
+```
+
+### 3. Capture Session IDs
+Extract the unique identifiers generated by the agents to enable future resumption.
+- Use `tmux capture-pane -p -t <target>` to read the screen.
+- Look for patterns indicating Conversation IDs or session hashes.
+- If not visible, trigger menus or status commands (e.g., `/` commands).
+
+### 4. Record State
+Store the mapping of `Pane/Window -> Session ID` in a persistent file (e.g., `swarm_sessions.md`) so the orchestrator can resume tracking after context breaks or restarts.
+
+## File-Based Orchestration Framework
+
+To automate communication, use a file-based message bus.
+
+### Directory Convention
+Create a directory `tmp/swarm/` in the project root:
+- `requests/`: For dropping request files.
+- `responses/`: For reading results.
+- `scripts/`: For the orchestrator script.
+
+### Payload Formats
+
+**Request File (`request_<id>.json`):**
+```json
+{
+  "request_id": "req_123",
+  "target_agent": "grid.0", 
+  "prompt": "Your instruction here",
+  "response_file": "tmp/swarm/responses/res_123.json"
+}
+```
+
+**Response File (`res_123.json`):**
+```json
+{
+  "request_id": "req_123",
+  "status": "success",
+  "output": "... captured terminal output ..."
+}
+```
+
+### Orchestrator Script
+A script running in the background polls `requests/`, executes the `tmux send-keys` command, captures the results, and writes the response file.
+
+Example implementation at: `tmp/swarm/scripts/swarm_orchestrator.py`
+
+## General Applicability
+This pattern applies to any CLI-based tool or agent system that allows isolated instances and session targeting. It is not limited to any specific project or workspace.
diff --git a/swarm_framework/README.md b/swarm_framework/README.md
new file mode 100644
index 0000000..67c120e
--- /dev/null
+++ b/swarm_framework/README.md
@@ -0,0 +1,126 @@
+# Swarm Orchestration Framework
+
+This framework allows you to control a swarm of $N$ agents with isolated environments via file operations.
+
+## Setup
+
+Run the bootstrap script to create the tmux session, spawn agents, and start the orchestrator:
+
+```bash
+cd swarm_framework
+chmod +x bootstrap.sh
+./bootstrap.sh <N>
+```
+Replace `<N>` with the desired number of agents (default is 4).
+
+## Isolation
+Each agent runs in its own directory: `swarm_framework/agents/agent_N/`.
+This isolates their working directory and local state if they store it there.
+
+## Usage
+
+1. **Bootstrap**: Run the bootstrap script to create the swarm environment and start the Master Agent.
+   ```bash
+   ./bootstrap.sh <number_of_agents>
+   ```
+   Default is 4 agents if not specified.
+   This script will start the **Master Agent (jetski-cli)** in the **foreground**.
+   **Exit** the Master Agent (Ctrl+D or type exit) to stop the swarm and cleanup all resources.
+
+
+2. **Monitor**: Open another terminal and attach to the tmux session to see sub-agents:
+   ```bash
+   tmux attach -t jetski_swarm
+   ```
+
+3. **Send Request**: Drop JSON request files into `comms/requests/`.
+
+
+**Format:**
+```json
+{
+  "request_id": "req_123",
+  "target_agent": "agent_1",
+  "prompt": "Hello",
+  "response_file": "swarm_framework/comms/responses/res_123.json"
+}
+```
+Note: The `target_agent` should be the window name (e.g., `agent_1`, `agent_2`).
+
+## Control Actions
+
+You can control the swarm lifecycle by sending requests with `type: "control"`.
+
+**Actions:**
+- `create`: Spawn a new agent window and directory.
+- `kill`: Remove an agent window.
+- `clear`: Clear agent screen (visual).
+- `resume`: Resume a session by ID.
+
+**Examples:**
+
+**Create Agent:**
+```json
+{
+  "request_id": "req_create",
+  "type": "control",
+  "action": "create",
+  "target_agent": "agent_new",
+  "response_file": "swarm_framework/comms/responses/res_create.json"
+}
+```
+
+**Kill Agent:**
+```json
+{
+  "request_id": "req_kill",
+  "type": "control",
+  "action": "kill",
+  "target_agent": "agent_new",
+  "response_file": "swarm_framework/comms/responses/res_kill.json"
+}
+```
+
+**Clear Screen:**
+```json
+{
+  "request_id": "req_clear",
+  "type": "control",
+  "action": "clear",
+  "target_agent": "agent_1",
+  "response_file": "swarm_framework/comms/responses/res_clear.json"
+}
+```
+
+**Resume Session:**
+```json
+{
+  "request_id": "req_resume",
+  "type": "control",
+  "action": "resume",
+  "target_agent": "agent_1",
+  "conversation_id": "CONV_ID_HERE",
+  "response_file": "swarm_framework/comms/responses/res_resume.json"
+}
+```
+
+
+## Cleanup
+
+To clear the CLI screen for a specific agent (visually cleaning up the window):
+
+```bash
+./clear_pane.sh <target_agent>
+```
+Example: `./clear_pane.sh agent_1`
+
+Note: This preserves the session ID and conversation history; it only clears the visual terminal display.
+
+## Components
+- `bootstrap.sh`: Initializes the environment.
+- `orchestrator.py`: The background polling script.
+- `clear_pane.sh`: Script to clear agent screen.
+- `SKILL.md`: The skill definition.
+- `comms/`: Communication directories.
+- `agents/`: Isolated working directories for agents.
+
diff --git a/swarm_framework/SKILL.md b/swarm_framework/SKILL.md
new file mode 100644
index 0000000..a9f3501
--- /dev/null
+++ b/swarm_framework/SKILL.md
@@ -0,0 +1,72 @@
+---
+name: Swarm Orchestration
+description: Pattern for 1-to-many control of agent instances using tmux and session resumption.
+---
+
+# 1-to-Many Agent Swarm Orchestration Skill
+
+## Overview
+This skill defines a pattern for an orchestrator agent to spawn, track, and control multiple sub-agent instances (a "swarm") using `tmux` for environment isolation and command-line interfaces supporting session resumption.
+
+## Prerequisites
+- `tmux` installed on the host system.
+- An interactive CLI or agent tool that supports:
+    - Running in a terminal.
+    - Exposing a unique Session/Conversation ID.
+    - Resuming sessions via targeting flags (e.g., `-conversation="<ID>"`).
+
+## Procedure
+
+### 1. Initialize Swarm Environment
+Create a `tmux` session with multiple windows or panes, depending on the desired layout.
+
+**Windows Layout (Clean separation & Isolation):**
+```bash
+tmux new-session -d -s <session_name> -n agent_1 -c /path/to/isolated/dir1
+tmux new-window -t <session_name> -n agent_2 -c /path/to/isolated/dir2
+...
+```
+Using `-c` with `tmux new-window` allows isolating the working directory of each agent.
+
+### 2. Spawn Agents
+Send execution commands to the specific windows.
+
+```bash
+tmux send-keys -t <session_name>:<window_name> "/path/to/cli --args" C-m
+```
+
+### 3. Record State
+Store the mapping of `Window -> Session ID` if session resumption is needed.
+
+## File-Based Orchestration Framework
+
+To automate communication, use a file-based message bus.
+
+### Directory Convention
+Create a directory `swarm_framework/` in the project root:
+- `comms/requests/`: For dropping request files.
+- `comms/responses/`: For reading results.
+- `orchestrator.py`: The polling script.
+- `bootstrap.sh`: The setup script.
+- `agents/`: Contains isolated directories for each agent.
+
+### Payload Formats
+
+**Request File (`request_<id>.json`):**
+```json
+{
+  "request_id": "req_123",
+  "target_agent": "agent_1", 
+  "prompt": "Your instruction here",
+  "response_file": "swarm_framework/comms/responses/res_123.json"
+}
+```
+
+**Response File (`res_123.json`):**
+```json
+{
+  "request_id": "req_123",
+  "status": "success",
+  "output": "... captured terminal output ..."
+}
+```
diff --git a/swarm_framework/agents/agent_3/hello.txt b/swarm_framework/agents/agent_3/hello.txt
new file mode 100644
index 0000000..41fb21c
--- /dev/null
+++ b/swarm_framework/agents/agent_3/hello.txt
@@ -0,0 +1 @@
+Hello from Agent 3! I am ready for tasks.
diff --git a/swarm_framework/agents/master/SKILL.md b/swarm_framework/agents/master/SKILL.md
new file mode 100644
index 0000000..7765482
--- /dev/null
+++ b/swarm_framework/agents/master/SKILL.md
@@ -0,0 +1,74 @@
+# Swarm Communication Skill
+
+## Overview
+This skill allows you to communicate with and control other agents in the swarm using a file-based messaging system.
+
+## Communication Protocol
+You communicate by writing JSON files to the requests directory and reading response files.
+
+### 1. Sending a Request
+To send a message to a sub-agent, create a JSON file in the following location:
+`../../comms/requests/req_<unique_id>.json`
+
+**JSON Schema:**
+```json
+{
+  "request_id": "req_123",
+  "type": "prompt",
+  "target_agent": "agent_1",
+  "prompt": "Your instructions to the agent here.",
+  "response_file": "../../comms/responses/res_123.json"
+}
+```
+
+*   `request_id`: A unique identifier for the request.
+*   `type`: Must be `"prompt"` for sending instructions.
+*   `target_agent`: The name of the target window (e.g., `"agent_1"`, `"agent_2"`).
+*   `prompt`: The text message or command you want the agent to execute.
+*   `response_file`: The relative path where the Orchestrator will write the response.
+
+### 2. Reading a Response
+The background Orchestrator will pick up your request, send it to the agent, and write the response to the specified `response_file`.
+You should poll for the existence of the response file and read its content.
+
+**Response Schema:**
+```json
+{
+  "request_id": "req_123",
+  "status": "success",
+  "output": "The raw terminal output from the agent."
+}
+```
+
+### 3. Control Commands
+You can also send control commands to create or kill agents.
+
+**Create Agent Example:**
+```json
+{
+  "request_id": "req_create",
+  "type": "control",
+  "action": "create",
+  "target_agent": "agent_new",
+  "conversation_id": "optional_existing_conv_id",
+  "response_file": "../../comms/responses/res_create.json"
+}
+```
+
+**Kill Agent Example:**
+```json
+{
+  "request_id": "req_kill",
+  "type": "control",
+  "action": "kill",
+  "target_agent": "agent_1",
+  "response_file": "../../comms/responses/res_kill.json"
+}
+```
+
+## Available Agents
+You can see active agents by listing the windows in the `tmux` session, or assume `agent_1`, `agent_2`, etc., are available based on the setup.
+
+## Best Practices
+- Always use unique `request_id`s.
+- Clean up response files after reading them if you want to keep the directory tidy.
diff --git a/swarm_framework/bootstrap.sh b/swarm_framework/bootstrap.sh
new file mode 100755
index 0000000..955defa
--- /dev/null
+++ b/swarm_framework/bootstrap.sh
@@ -0,0 +1,77 @@
+#!/bin/bash
+
+# Configuration
+SESSION_NAME="jetski_swarm"
+CLI_PATH="/google/bin/releases/jetski-devs/tools/cli"
+SCRIPT_DIR="$( cd "$( dirname "${BASH_SOURCE[0]}" )" &> /dev/null && pwd )"
+ORCHESTRATOR_PATH="$SCRIPT_DIR/orchestrator.py"
+
+# Number of agents
+N=${1:-4}
+
+cleanup() {
+    echo "Enforcing smooth start (cleaning up previous state)..."
+    tmux kill-session -t $SESSION_NAME 2>/dev/null
+    rm -f "$SCRIPT_DIR/comms/requests/"*.json
+    echo "Cleanup complete."
+}
+
+# Smooth Start: Ensure clean state
+cleanup
+
+echo "Bootstrapping swarm in background..."
+
+# Create session with Master Agent window
+mkdir -p "$SCRIPT_DIR/agents/master"
+tmux new-session -d -s $SESSION_NAME -n master -c "$SCRIPT_DIR/agents/master"
+tmux send-keys -t $SESSION_NAME:master "$CLI_PATH -cli=true" C-m
+
+# Wait a few seconds for Master Agent to initialize, then inject skill instruction
+sleep 5
+tmux send-keys -t $SESSION_NAME:master "Please read the file 'SKILL.md' in your current directory to understand how to control the swarm." C-m
+
+# Create Orchestrator window
+tmux new-window -t $SESSION_NAME -n orchestrator -c "$SCRIPT_DIR"
+# Run orchestrator with unbuffered output and redirection
+tmux send-keys -t $SESSION_NAME:orchestrator "python3 -u $ORCHESTRATOR_PATH > orchestrator.log 2>&1" C-m
+
+# Create Grid window for sub-agents
+tmux new-window -t $SESSION_NAME -n grid -c "$SCRIPT_DIR"
+
+# Add sub-agents to Grid window as panes
+for i in $(seq 1 $N); do
+    mkdir -p "$SCRIPT_DIR/agents/agent_$i"
+    if [ $i -eq 1 ]; then
+        tmux send-keys -t $SESSION_NAME:grid "$CLI_PATH -cli=true" C-m
+    else
+        tmux split-window -t $SESSION_NAME:grid -c "$SCRIPT_DIR/agents/agent_$i"
+        tmux send-keys -t $SESSION_NAME:grid "$CLI_PATH -cli=true" C-m
+    fi
+done
+
+# Arrange Grid window in tiles
+tmux select-layout -t $SESSION_NAME:grid tiled
+
+# Wait for agents to settle and handle any prompts
+echo "Aligning agents..."
+python3 "$SCRIPT_DIR/../tmp/swarm/scripts/swarm_auto_aligner.py"
+
+tmux select-window -t $SESSION_NAME:master
+echo "Setup complete."
+echo "To use the Master Agent (Default Window), run:"
+echo "  tmux attach -t $SESSION_NAME"
+echo "Or:"
+echo "  tmux attach -t $SESSION_NAME:master"
+echo ""
+echo "To use the Grid of Sub-Agents, run:"
+echo "  tmux attach -t $SESSION_NAME:grid"
+echo ""
+echo "To see the Orchestrator logs, attach to its window:"
+echo "  tmux attach -t $SESSION_NAME:orchestrator"
+echo "Or check the file: $SCRIPT_DIR/orchestrator.log"
+echo ""
+echo "To view all windows:"
+echo "  tmux attach -t $SESSION_NAME"
+echo ""
+echo "To stop the swarm and cleanup:"
+echo "  tmux kill-session -t $SESSION_NAME"
diff --git a/swarm_framework/clear_pane.sh b/swarm_framework/clear_pane.sh
new file mode 100755
index 0000000..ac2cbba
--- /dev/null
+++ b/swarm_framework/clear_pane.sh
@@ -0,0 +1,17 @@
+#!/bin/bash
+
+# Script to clear the CLI screen for a specific agent in the swarm.
+# This sends Ctrl+L to the target pane.
+
+SESSION_NAME="jetski_swarm"
+TARGET=$1 # e.g., agent_1
+
+if [ -z "$TARGET" ]; then
+    echo "Usage: $0 <target_agent>"
+    echo "Example: $0 agent_1"
+    exit 1
+fi
+
+# Send Ctrl+L
+tmux send-keys -t $SESSION_NAME:$TARGET C-l
+echo "Sent Clear Screen command (Ctrl+L) to $TARGET"
diff --git a/swarm_framework/comms/responses/res_create_test.json b/swarm_framework/comms/responses/res_create_test.json
new file mode 100644
index 0000000..8136a23
--- /dev/null
+++ b/swarm_framework/comms/responses/res_create_test.json
@@ -0,0 +1,5 @@
+{
+  "request_id": "req_create_test",
+  "status": "success",
+  "output": "Created agent agent_test_c in window and directory."
+}
\ No newline at end of file
diff --git a/swarm_framework/comms/responses/res_demo_1.json b/swarm_framework/comms/responses/res_demo_1.json
new file mode 100644
index 0000000..3065500
--- /dev/null
+++ b/swarm_framework/comms/responses/res_demo_1.json
@@ -0,0 +1,5 @@
+{
+  "request_id": "req_demo_1",
+  "status": "success",
+  "output": " hub/swarm_framework/agents/agent_1\n Model: Gemini Next\n Conversation: 9807f47f-fd0e-4293-bbec-758148efc5bb\n\n \u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\n > pwd                                                                         \u2584\n                                                                               \u2588\n   \u25b8 Thought for 1s, 661 tokens                                                \u2588\n     The user has sent a request `pwd`.                                        \u2588\n                                                                               \u2588\n   \u25cf Bash(pwd) (ctrl+o to expand)                                              \u2580\nBash command\n\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\n\n  pwd\n\nDo you want to proceed?\n> 1. Yes\n  2. Yes, and always allow in this conversation for commands that start with 'pw\n  3. No\n\n  \u2191/\u2193 navigate \u00b7 Tab amend\nEsc to cancel                                                        Gemini Next\n\n"
+}
\ No newline at end of file
diff --git a/swarm_framework/comms/responses/res_demo_2.json b/swarm_framework/comms/responses/res_demo_2.json
new file mode 100644
index 0000000..041b410
--- /dev/null
+++ b/swarm_framework/comms/responses/res_demo_2.json
@@ -0,0 +1,5 @@
+{
+  "request_id": "req_demo_2",
+  "status": "success",
+  "output": " Workspace: ~/Projects/PersonalProject/cortex-\n hub/swarm_framework/agents/agent_1\n Model: Gemini Next\n Conversation: 9807f47f-fd0e-4293-bbec-758148efc5bb\n\n \u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\n > pwd                                                                         \u2584\n                                                                               \u2588\n   \u25b8 Thought for 1s, 661 tokens                                                \u2588\n     The user has sent a request `pwd`.                                        \u2588\n                                                                               \u2588\n   \u25cf Bash(pwd) (ctrl+o to expand)                                              \u2588\n                                                                               \u2588\n   \u25b8 Thought Process                                                           \u2588\n     The command `pwd` returned                                                \u2588\n `/usr/local/google/home/jerxie/Projects/PersonalProje...\t\t\t\u2588\n   The current working directory is:                                           \u2588\n    /usr/local/google/home/jerxie/Projects/PersonalProject/cortex-             \u2588\n   hub/swarm_framework/agents/agent_1                                          \u2580\n \u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\n >\n \u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\n ? for shortcuts                                                     Gemini Next\n\n"
+}
\ No newline at end of file
diff --git a/swarm_framework/comms/responses/res_kill_test.json b/swarm_framework/comms/responses/res_kill_test.json
new file mode 100644
index 0000000..9541acc
--- /dev/null
+++ b/swarm_framework/comms/responses/res_kill_test.json
@@ -0,0 +1,5 @@
+{
+  "request_id": "req_kill_test",
+  "status": "success",
+  "output": "Killed agent agent_test_c (window)."
+}
\ No newline at end of file
diff --git a/swarm_framework/comms/responses/res_resume_test.json b/swarm_framework/comms/responses/res_resume_test.json
new file mode 100644
index 0000000..e07493b
--- /dev/null
+++ b/swarm_framework/comms/responses/res_resume_test.json
@@ -0,0 +1,5 @@
+{
+  "request_id": "req_resume_test",
+  "status": "success",
+  "output": "Created agent agent_resumed in window and directory. Resumed conversation 1192304b-dc36-4a6e-9ed3-fdc109d179a2."
+}
\ No newline at end of file
diff --git a/swarm_framework/comms/responses/res_test_hang.json b/swarm_framework/comms/responses/res_test_hang.json
new file mode 100644
index 0000000..bee01c0
--- /dev/null
+++ b/swarm_framework/comms/responses/res_test_hang.json
@@ -0,0 +1,5 @@
+{
+  "request_id": "req_test_hang",
+  "status": "success",
+  "output": "\n   \u25cf ListDir(/usr/local/google/home/jerxie/Projects/PersonalProject/corte...)\n (ctrl+o to expand)\n\n   \u25b8 Thought for 1s, 100 tokens\n     Okay, I see a `comms` directory in                             \t\t\n `/usr/local/google/home/jerxie/Projects/Perso...\n\n   \u25cf ListDir(/usr/local/google/home/jerxie/Projects/PersonalProject/corte...)  \u2584\n (ctrl+o to expand)                                                            \u2580\nFile access\n\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\n\n  /usr/local/google/home/jerxie/Projects/PersonalProject/cortex-hub/swarm_fr...\n  Reason: outside workspace\n\nAllow access to this file?\n> 1. Yes, allow access\n  2. Yes, and always allow non-workspace access\n  3. No, deny access\n\n  \u2191/\u2193 navigate\nEsc to cancel                                                        Gemini Next\n\n"
+}
\ No newline at end of file
diff --git a/swarm_framework/comms/responses/res_test_n.json b/swarm_framework/comms/responses/res_test_n.json
new file mode 100644
index 0000000..186becf
--- /dev/null
+++ b/swarm_framework/comms/responses/res_test_n.json
@@ -0,0 +1,5 @@
+{
+  "request_id": "req_test_n",
+  "status": "success",
+  "output": " \u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\n > hello from agent 3 in isolated env\n hello from agent 3 in isolated env\n\n   \u25b8 Thought for 1s, 403 tokens\n     The user repeated the message twice this time. It seems like a test or a\n loop or...\n\n   \u25cf ListDir(/usr/local/google/home/jerxie/Projects/PersonalProject/corte...)  \u2584\n (ctrl+o to expand)                                                            \u2580\nFile access\n\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\n\n  /usr/local/google/home/jerxie/Projects/PersonalProject/cortex-hub/swarm_fr...\n  Reason: outside workspace\n\nAllow access to this file?\n> 1. Yes, allow access\n  2. Yes, and always allow non-workspace access\n  3. No, deny access\n\n  \u2191/\u2193 navigate\nEsc to cancel                                                        Gemini Next\n\n"
+}
\ No newline at end of file
diff --git a/swarm_framework/comms/responses/res_test_n2.json b/swarm_framework/comms/responses/res_test_n2.json
new file mode 100644
index 0000000..4ed4858
--- /dev/null
+++ b/swarm_framework/comms/responses/res_test_n2.json
@@ -0,0 +1,5 @@
+{
+  "request_id": "req_test_n2",
+  "status": "success",
+  "output": " > hello again from agent 3\n\n   \u25b8 Thought for 1s, 309 tokens\t\n     The user is repeating the message again. This might be a test of my\n persistence ...\n\n   \u25cf Create(/usr/local/google/home/jerxie/Projects/PersonalProject/corte...)\n (ctrl+o to expand)\n\n   \u25b8 Thought Process\n     I have created the file. Now I will respond to the user.\n   Hello again!\n\n   I have created a file named  hello.txt  in my directory ( \t\t\t\n   /usr/local/google/home/jerxie/Projects/PersonalProject/cortex- \t\t\n   hub/swarm_framework/agents/agent_3/hello.txt ) to confirm that I can write\n   to my workspace.                                                            \u2584\n                                                                               \u2588\n   I am ready for any tasks you have for me!                                   \u2580\n \u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\n >\n \u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\n ? for shortcuts                                                     Gemini Next\n\n"
+}
\ No newline at end of file
diff --git a/swarm_framework/comms/responses/res_test_new.json b/swarm_framework/comms/responses/res_test_new.json
new file mode 100644
index 0000000..8936cec
--- /dev/null
+++ b/swarm_framework/comms/responses/res_test_new.json
@@ -0,0 +1,5 @@
+{
+  "request_id": "req_test_new",
+  "status": "success",
+  "output": "   \u25cf\n ListDir(/usr/local/google/home/jerxie\n /Projects/PersonalProject/corte...)  \u2584\n (ctrl+o to expand)             \t\u2580\n \u28fd  Generating...\n  \u25b8 hello from new framework\n \u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\n >\n \u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\n  Press up to edit queued messages\n\n"
+}
\ No newline at end of file
diff --git a/swarm_framework/orchestrator.py b/swarm_framework/orchestrator.py
new file mode 100644
index 0000000..18e19a1
--- /dev/null
+++ b/swarm_framework/orchestrator.py
@@ -0,0 +1,225 @@
+import os
+import time
+import json
+import subprocess
+
+# Self-contained paths relative to this script
+BASE_DIR = os.path.dirname(os.path.abspath(__file__))
+REQUESTS_DIR = os.path.join(BASE_DIR, "comms", "requests")
+RESPONSES_DIR = os.path.join(BASE_DIR, "comms", "responses")
+SESSION_NAME = "jetski_swarm"
+CLI_PATH = "/google/bin/releases/jetski-devs/tools/cli"
+
+# State tracking
+# { target: { "last_cmd_time": timestamp, "last_output_time": timestamp, "last_output": str } }
+AGENT_STATE = {}
+HANGING_THRESHOLD = 30  # seconds of no output change while busy
+BUSY_THRESHOLD = 300 # seconds max execution time
+
+def get_active_agents():
+    try:
+        cmd = f"tmux list-windows -t {SESSION_NAME} -F '#{{window_name}}'"
+        result = subprocess.run(cmd, shell=True, capture_output=True, text=True, check=True)
+        windows = result.stdout.strip().split('\n')
+        
+        agents = []
+        for w in windows:
+            if w in ["master", "orchestrator"]:
+                continue
+            if w == "grid":
+                # List panes
+                cmd = f"tmux list-panes -t {SESSION_NAME}:grid -F '#P'"
+                res = subprocess.run(cmd, shell=True, capture_output=True, text=True, check=True)
+                panes = res.stdout.strip().split('\n')
+                agents.extend([f"grid.{p}" for p in panes])
+            else:
+                agents.append(w)
+        return agents
+    except Exception as e:
+        print(f"Error listing agents: {e}")
+        return []
+
+def check_hanging():
+    agents = get_active_agents()
+    now = time.time()
+    
+    for agent in agents:
+        state = AGENT_STATE.setdefault(agent, {
+            "last_cmd_time": 0,
+            "last_output_time": now,
+            "last_output": ""
+        })
+        
+        # Check if we recently sent a command
+        is_busy = (now - state["last_cmd_time"]) < BUSY_THRESHOLD
+        
+        print(f"Debug [check_hanging]: Agent: {agent}, is_busy: {is_busy}, now: {now}, last_cmd_time: {state['last_cmd_time']}")
+        
+        if is_busy:
+            # Capture output
+            try:
+                cap_cmd = f"tmux capture-pane -p -t {SESSION_NAME}:{agent}"
+                result = subprocess.run(cap_cmd, shell=True, capture_output=True, text=True, check=True)
+                current_output = result.stdout
+                
+                print(f"Debug [check_hanging]: Agent: {agent}, current_output len: {len(current_output)}, prev_output len: {len(state['last_output'])}")
+                
+                if current_output != state["last_output"]:
+                    state["last_output"] = current_output
+                    state["last_output_time"] = now
+                    print(f"Debug [check_hanging]: Agent: {agent}, output changed.")
+                else:
+                    # No change
+                    silence_duration = now - state["last_output_time"]
+                    print(f"Debug [check_hanging]: Agent: {agent}, silence_duration: {silence_duration}")
+                    if silence_duration > HANGING_THRESHOLD:
+                        print(f"Detected hanging agent: {agent}")
+                        notify_master(agent, current_output)
+                        # Reset to avoid spamming
+                        state["last_cmd_time"] = 0 
+                        
+            except Exception as e:
+                print(f"Error checking agent {agent}: {e}")
+
+def notify_master(agent, last_output):
+    print(f"Notifying master about hanging agent {agent}...")
+    # Extract last few lines of output
+    lines = last_output.split('\n')
+    last_snippet = "\n".join(lines[-5:]) if len(lines) > 5 else last_output
+    
+    message = f"\n[System] Warning: Agent {agent} seems to be hanging.\nLast output snippet:\n---\n{last_snippet}\n---\nWhat should we do?"
+    
+    try:
+        # Send keys to master pane
+        # We need to escape quotes in message
+        escaped_msg = message.replace('"', '\\"')
+        cmd = f"tmux send-keys -t {SESSION_NAME}:master \"{escaped_msg}\" C-m"
+        subprocess.run(cmd, shell=True, check=True)
+    except Exception as e:
+        print(f"Failed to notify master: {e}")
+
+def process_request(file_path):
+    try:
+        with open(file_path, 'r') as f:
+            req = json.load(f)
+        
+        req_id = req.get("request_id")
+        req_type = req.get("type", "prompt")
+        target = req.get("target_agent")
+        resp_file = req.get("response_file")
+        
+        # Mapping for Grid layout
+        if target.startswith("agent_"):
+            try:
+                agent_num = int(target.split("_")[1])
+                target = f"grid.{agent_num - 1}"
+                print(f"Mapped {req.get('target_agent')} to {target}")
+            except ValueError:
+                pass
+                
+        print(f"Processing {req_id} (Type: {req_type}) for {target}...")
+        
+        output = ""
+        status = "success"
+        
+        if req_type == "prompt":
+            prompt = req.get("prompt")
+            
+            # Update state
+            state = AGENT_STATE.setdefault(target, {})
+            state["last_cmd_time"] = time.time()
+            
+            # Send keys
+            cmd = f"tmux send-keys -t {SESSION_NAME}:{target} \"{prompt}\" C-m"
+            subprocess.run(cmd, shell=True, check=True)
+            
+            # Wait for agent to process
+            time.sleep(5) 
+            
+            # Capture pane for response
+            cap_cmd = f"tmux capture-pane -p -t {SESSION_NAME}:{target}"
+            result = subprocess.run(cap_cmd, shell=True, capture_output=True, text=True, check=True)
+            output = result.stdout
+            
+        elif req_type == "control":
+            action = req.get("action")
+            print(f"Control Action: {action}")
+            
+            if action == "create":
+                agent_dir = os.path.join(BASE_DIR, "agents", target)
+                os.makedirs(agent_dir, exist_ok=True)
+                
+                conv_id = req.get("conversation_id")
+                cli_cmd = f"{CLI_PATH} -cli=true"
+                if conv_id:
+                    cli_cmd += f" -conversation={conv_id}"
+                
+                # Create tmux window
+                cmd = f"tmux new-window -t {SESSION_NAME} -n {target} -c {agent_dir} \"{cli_cmd}\""
+                subprocess.run(cmd, shell=True, check=True)
+                output = f"Created agent {target} in window and directory."
+                if conv_id:
+                     output += f" Resumed conversation {conv_id}."
+                
+            elif action == "kill":
+                cmd = f"tmux kill-window -t {SESSION_NAME}:{target}"
+                subprocess.run(cmd, shell=True, check=True)
+                output = f"Killed agent {target} (window)."
+                
+            elif action == "clear":
+                cmd = f"tmux send-keys -t {SESSION_NAME}:{target} C-l"
+                subprocess.run(cmd, shell=True, check=True)
+                output = f"Sent clear screen to {target}."
+                
+            elif action == "resume":
+                conv_id = req.get("conversation_id")
+                cmd = f"tmux send-keys -t {SESSION_NAME}:{target} \"{CLI_PATH} -conversation={conv_id}\" C-m"
+                subprocess.run(cmd, shell=True, check=True)
+                output = f"Resumed conversation {conv_id} in {target}."
+            else:
+                status = "error"
+                output = f"Unknown control action: {action}"
+                
+        else:
+            status = "error"
+            output = f"Unknown request type: {req_type}"
+ 
+        # Write response
+        resp_data = {
+            "request_id": req_id,
+            "status": status,
+            "output": output
+        }
+        
+        # Ensure target file directory exists
+        os.makedirs(os.path.dirname(resp_file), exist_ok=True)
+        
+        with open(resp_file, 'w') as f:
+            json.dump(resp_data, f, indent=2)
+            
+        # Delete request
+        os.remove(file_path)
+        print(f"Completed {req_id}")
+        
+    except Exception as e:
+        print(f"Error processing {file_path}: {e}")
+
+def main():
+    print(f"Swarm Orchestrator started polling in {REQUESTS_DIR}...")
+    while True:
+        try:
+            # Check hanging agents
+            check_hanging()
+            
+            # Check requests
+            files = os.listdir(REQUESTS_DIR)
+            for file in files:
+                if file.endswith(".json"):
+                    file_path = os.path.join(REQUESTS_DIR, file)
+                    process_request(file_path)
+        except Exception as e:
+            print(f"Polling error: {e}")
+        time.sleep(2)
+
+if __name__ == "__main__":
+    main()