diff --git a/.github/workflows/dev.lock.yml b/.github/workflows/dev.lock.yml
index 93b1d8c261..2db75f9b0d 100644
--- a/.github/workflows/dev.lock.yml
+++ b/.github/workflows/dev.lock.yml
@@ -2,7 +2,7 @@
 # To update this file, edit the corresponding .md file and run:
 #   gh aw compile
 
-name: "Dev"
+name: "dev"
 on:
   push:
     branches:
@@ -15,7 +15,7 @@ permissions: {}
 concurrency:
   group: "gh-aw-${{ github.workflow }}-${{ github.ref }}"
 
-run-name: "Dev"
+run-name: "dev"
 
 jobs:
   task:
@@ -120,6 +120,26 @@ jobs:
     steps:
       - name: Checkout repository
         uses: actions/checkout@v5
+      # Cache memory file share configuration from frontmatter processed below
+      - name: Create cache-memory directory
+        run: |
+          mkdir -p /tmp/cache-memory
+          echo "Cache memory directory created at /tmp/cache-memory"
+          echo "This folder provides persistent file storage across workflow runs"
+          echo "LLMs and agentic tools can freely read and write files in this directory"
+      - name: Cache memory file share data
+        uses: actions/cache@v4
+        with:
+          key: memory-${{ github.workflow }}-${{ github.run_id }}
+          path: /tmp/cache-memory
+          restore-keys: |
+            memory-${{ github.workflow }}-
+            memory-
+      - name: Upload cache-memory data as artifact
+        uses: actions/upload-artifact@v4
+        with:
+          name: cache-memory
+          path: /tmp/cache-memory
       - name: Generate Claude Settings
         run: |
           mkdir -p /tmp/.claude
@@ -724,8 +744,49 @@ jobs:
         run: |
           mkdir -p /tmp/aw-prompts
           cat > $GITHUB_AW_PROMPT << 'EOF'
+          # Development Assistant
+          
+          You are a development assistant that helps with coding tasks and maintains an execution plan.
+          
+          ## Task
+          
           Try to call a tool, `draw_pelican` that draws a pelican.
           
+          ## Execution Plan Management
+          
+          - **Check for existing plan**: Look for `/tmp/cache-memory/plan.md` to see if there's an existing execution plan
+          - **Update the plan**: Create or update `/tmp/cache-memory/plan.md` with your execution strategy
+          - **Remember progress**: Store any important findings, decisions, or progress in the cache folder
+          - **Recall previous work**: If a plan exists, reference it and continue from where you left off
+          
+          The plan should include:
+          1. Current objectives
+          2. Steps completed
+          3. Next steps
+          4. Any important discoveries or decisions
+          5. Tools or approaches that worked well
+          
+          Please maintain this execution plan throughout your work to ensure continuity across workflow runs.
+          
+          
+          ---
+          
+          ## Cache Folder Available
+          
+          You have access to a persistent cache folder at `/tmp/cache-memory/` where you can read and write files to create memories and store information.
+          
+          - **Read/Write Access**: You can freely read from and write to any files in this folder
+          - **Persistence**: Files in this folder persist across workflow runs via GitHub Actions cache
+          - **Last Write Wins**: If multiple processes write to the same file, the last write will be preserved
+          - **File Share**: Use this as a simple file share - organize files as you see fit
+          
+          Examples of what you can store:
+          - `/tmp/cache-memory/notes.txt` - general notes and observations
+          - `/tmp/cache-memory/preferences.json` - user preferences and settings
+          - `/tmp/cache-memory/history.log` - activity history and logs
+          - `/tmp/cache-memory/state/` - organized state files in subdirectories
+          
+          Feel free to create, read, update, and organize files in this folder as needed for your tasks.
           
           ---
           
@@ -753,7 +814,7 @@ jobs:
               engine_name: "Claude Code",
               model: "",
               version: "",
-              workflow_name: "Dev",
+              workflow_name: "dev",
               experimental: false,
               supports_tools_whitelist: true,
               supports_http_transport: true,
@@ -784,15 +845,19 @@ jobs:
       - name: Execute Claude Code CLI
         id: agentic_execution
         # Allowed tools (sorted):
+        # - Edit(/tmp/cache-memory/*)
         # - ExitPlanMode
         # - Glob
         # - Grep
         # - LS
+        # - MultiEdit(/tmp/cache-memory/*)
         # - NotebookRead
         # - Read
+        # - Read(/tmp/cache-memory/*)
         # - Task
         # - TodoWrite
         # - Write
+        # - Write(/tmp/cache-memory/*)
         # - mcp__github__download_workflow_run_artifact
         # - mcp__github__get_code_scanning_alert
         # - mcp__github__get_commit
@@ -841,7 +906,7 @@ jobs:
         run: |
           set -o pipefail
           # Execute Claude Code CLI with prompt from file
-          npx @anthropic-ai/claude-code@latest --print --max-turns 5 --mcp-config /tmp/mcp-config/mcp-servers.json --allowed-tools "ExitPlanMode,Glob,Grep,LS,NotebookRead,Read,Task,TodoWrite,Write,mcp__github__download_workflow_run_artifact,mcp__github__get_code_scanning_alert,mcp__github__get_commit,mcp__github__get_dependabot_alert,mcp__github__get_discussion,mcp__github__get_discussion_comments,mcp__github__get_file_contents,mcp__github__get_issue,mcp__github__get_issue_comments,mcp__github__get_job_logs,mcp__github__get_me,mcp__github__get_notification_details,mcp__github__get_pull_request,mcp__github__get_pull_request_comments,mcp__github__get_pull_request_diff,mcp__github__get_pull_request_files,mcp__github__get_pull_request_reviews,mcp__github__get_pull_request_status,mcp__github__get_secret_scanning_alert,mcp__github__get_tag,mcp__github__get_workflow_run,mcp__github__get_workflow_run_logs,mcp__github__get_workflow_run_usage,mcp__github__list_branches,mcp__github__list_code_scanning_alerts,mcp__github__list_commits,mcp__github__list_dependabot_alerts,mcp__github__list_discussion_categories,mcp__github__list_discussions,mcp__github__list_issues,mcp__github__list_notifications,mcp__github__list_pull_requests,mcp__github__list_secret_scanning_alerts,mcp__github__list_tags,mcp__github__list_workflow_jobs,mcp__github__list_workflow_run_artifacts,mcp__github__list_workflow_runs,mcp__github__list_workflows,mcp__github__search_code,mcp__github__search_issues,mcp__github__search_orgs,mcp__github__search_pull_requests,mcp__github__search_repositories,mcp__github__search_users" --debug --verbose --permission-mode bypassPermissions --output-format json --settings /tmp/.claude/settings.json "$(cat /tmp/aw-prompts/prompt.txt)" 2>&1 | tee /tmp/dev.log
+          npx @anthropic-ai/claude-code@latest --print --max-turns 5 --mcp-config /tmp/mcp-config/mcp-servers.json --allowed-tools "Edit(/tmp/cache-memory/*),ExitPlanMode,Glob,Grep,LS,MultiEdit(/tmp/cache-memory/*),NotebookRead,Read,Read(/tmp/cache-memory/*),Task,TodoWrite,Write,Write(/tmp/cache-memory/*),mcp__github__download_workflow_run_artifact,mcp__github__get_code_scanning_alert,mcp__github__get_commit,mcp__github__get_dependabot_alert,mcp__github__get_discussion,mcp__github__get_discussion_comments,mcp__github__get_file_contents,mcp__github__get_issue,mcp__github__get_issue_comments,mcp__github__get_job_logs,mcp__github__get_me,mcp__github__get_notification_details,mcp__github__get_pull_request,mcp__github__get_pull_request_comments,mcp__github__get_pull_request_diff,mcp__github__get_pull_request_files,mcp__github__get_pull_request_reviews,mcp__github__get_pull_request_status,mcp__github__get_secret_scanning_alert,mcp__github__get_tag,mcp__github__get_workflow_run,mcp__github__get_workflow_run_logs,mcp__github__get_workflow_run_usage,mcp__github__list_branches,mcp__github__list_code_scanning_alerts,mcp__github__list_commits,mcp__github__list_dependabot_alerts,mcp__github__list_discussion_categories,mcp__github__list_discussions,mcp__github__list_issues,mcp__github__list_notifications,mcp__github__list_pull_requests,mcp__github__list_secret_scanning_alerts,mcp__github__list_tags,mcp__github__list_workflow_jobs,mcp__github__list_workflow_run_artifacts,mcp__github__list_workflow_runs,mcp__github__list_workflows,mcp__github__search_code,mcp__github__search_issues,mcp__github__search_orgs,mcp__github__search_pull_requests,mcp__github__search_repositories,mcp__github__search_users" --debug --verbose --permission-mode bypassPermissions --output-format json --settings /tmp/.claude/settings.json "$(cat /tmp/aw-prompts/prompt.txt)" 2>&1 | tee /tmp/dev.log
         env:
           ANTHROPIC_API_KEY: ${{ secrets.ANTHROPIC_API_KEY }}
           DISABLE_TELEMETRY: "1"
diff --git a/.github/workflows/dev.md b/.github/workflows/dev.md
index 36f87303e2..7ae5049597 100644
--- a/.github/workflows/dev.md
+++ b/.github/workflows/dev.md
@@ -1,10 +1,13 @@
 ---
+name: "dev"
 on:
   workflow_dispatch: # do not remove this trigger
   push:
     branches:
       - copilot/*
       - pelikhan/*
+tools:
+  cache-memory: true
 safe-outputs:
   missing-tool:
   staged: true
@@ -14,4 +17,26 @@ engine:
 permissions: read-all
 ---
 
-Try to call a tool, `draw_pelican` that draws a pelican.
\ No newline at end of file
+# Development Assistant
+
+You are a development assistant that helps with coding tasks and maintains an execution plan.
+
+## Task
+
+Try to call a tool, `draw_pelican` that draws a pelican.
+
+## Execution Plan Management
+
+- **Check for existing plan**: Look for `/tmp/cache-memory/plan.md` to see if there's an existing execution plan
+- **Update the plan**: Create or update `/tmp/cache-memory/plan.md` with your execution strategy
+- **Remember progress**: Store any important findings, decisions, or progress in the cache folder
+- **Recall previous work**: If a plan exists, reference it and continue from where you left off
+
+The plan should include:
+1. Current objectives
+2. Steps completed
+3. Next steps
+4. Any important discoveries or decisions
+5. Tools or approaches that worked well
+
+Please maintain this execution plan throughout your work to ensure continuity across workflow runs.
\ No newline at end of file
diff --git a/docs/src/content/docs/reference/cache-memory.md b/docs/src/content/docs/reference/cache-memory.md
index e9014dc3c4..c9273ab831 100644
--- a/docs/src/content/docs/reference/cache-memory.md
+++ b/docs/src/content/docs/reference/cache-memory.md
@@ -1,17 +1,17 @@
 ---
 title: Cache Memory
-description: Complete guide to using cache-memory for persistent memory across workflow runs using GitHub Actions cache and MCP memory servers.
+description: Complete guide to using cache-memory for persistent file storage across workflow runs using GitHub Actions cache and simple file operations.
 ---
 
-The `cache-memory` feature enables agentic workflows to maintain persistent memory across workflow runs by integrating the Model Context Protocol (MCP) memory server with GitHub Actions cache.
+The `cache-memory` feature enables agentic workflows to maintain persistent file storage across workflow runs using GitHub Actions cache infrastructure and simple file operations.
 
 ## Overview
 
 Cache Memory provides:
 
-- **Persistent Memory**: AI agents can remember information across multiple workflow runs
+- **Persistent File Storage**: AI agents can store and retrieve files across multiple workflow runs
 - **GitHub Actions Integration**: Built on top of GitHub Actions cache infrastructure 
-- **MCP Memory Server**: Uses the official Model Context Protocol memory server via npx
+- **Simple File Operations**: Uses standard file system operations instead of specialized tools
 - **Automatic Configuration**: Seamlessly integrates with Claude and Custom engines
 - **Smart Caching**: Intelligent cache key generation and restoration strategies
 
@@ -19,11 +19,11 @@ Cache Memory provides:
 
 When `cache-memory` is enabled, the workflow compiler automatically:
 
-1. **Mounts Memory MCP Server**: Configures an official MCP memory server via npx
-2. **Creates Cache Steps**: Adds GitHub Actions cache steps to restore and save memory data
-3. **Persistent Storage**: Maps `/tmp/cache-memory` to store memory data files
+1. **Creates Cache Directory**: Sets up `/tmp/cache-memory/` directory for file storage
+2. **Creates Cache Steps**: Adds GitHub Actions cache steps to restore and save data
+3. **Persistent Storage**: Maps `/tmp/cache-memory/` to store user data files
 4. **Cache Key Management**: Generates intelligent cache keys with progressive fallback
-5. **Tool Integration**: Adds "memory" tool to the MCP configuration for AI engines
+5. **Prompts LLM**: Informs the AI agent about the available cache folder
 
 ## Enabling Cache Memory
 
@@ -41,42 +41,40 @@ tools:
 
 This uses:
 - **Default cache key**: `memory-${{ github.workflow }}-${{ github.run_id }}`
-- **Default setup**: Uses `npx @modelcontextprotocol/server-memory` 
-- **Default storage path**: `/tmp/cache-memory` for memory data files
+- **Simple file access**: Uses standard file operations for read/write
+- **Default storage path**: `/tmp/cache-memory/` for data files
 
-## Prompting for Memory Operations
+## Using the Cache Folder
 
-Memory is accessed via the `memory` tool in your workflow steps.
+The cache folder is accessible to AI agents and provides file system access for persistent storage.
 
 ### Storing Information
 
-The AI agent can store information using the memory tool:
+The AI agent can store information using standard file operations:
 
 ```
-Remember that the user prefers verbose error messages when debugging.
+Please save this information to a file in the cache folder: "User prefers verbose error messages when debugging."
 ```
 
 ### Retrieving Information
 
-The AI agent can query its memory:
+The AI agent can read files from the cache folder:
 
 ```
-What do I know about the user's debugging preferences?
+Check what information I have stored in the cache folder from previous runs.
 ```
 
-### Memory Categories
+### File Organization
 
-The MCP memory server organizes information into categories:
+The AI agent can organize files as needed:
 
-- **Basic Identity**: User identification and context
-- **Behaviors**: Observed patterns and preferences  
-- **Preferences**: Explicit user preferences
-- **Goals**: Stated objectives and tasks
-- **Relationships**: Connections between entities
+- **Structured data**: JSON or YAML files for configuration and preferences
+- **Text files**: Plain text for notes, logs, and observations
+- **Directories**: Organize files into subdirectories for better structure
 
 ## Advanced Configuration
 
-You can also use the more advanced configuration to customize cache key and artifact retention:
+You can customize cache key and artifact retention:
 
 ```yaml
 ---
@@ -92,7 +90,7 @@ tools:
 
 ### Artifact Retention
 
-Configure how long memory data artifacts are retained:
+Configure how long cache data artifacts are retained:
 
 ```yaml
 ---
@@ -109,7 +107,7 @@ tools:
 The `retention-days` option controls the `actions/upload-artifact` retention period:
 - **Range**: 1-90 days
 - **Default**: Repository setting (if not specified)
-- **Purpose**: Provides alternative access to memory data beyond cache expiration
+- **Purpose**: Provides alternative access to cache data beyond cache expiration
 
 ## Cache Behavior and GitHub Actions Integration
 
@@ -121,10 +119,10 @@ Cache Memory leverages GitHub Actions cache with these characteristics:
 - **LRU Eviction**: Least recently used caches are evicted when limits are reached
 
 #### Artifact Upload (Optional)
-When `retention-days` is configured, memory data is also uploaded as artifacts:
+When `retention-days` is configured, cache data is also uploaded as artifacts:
 - **Retention Period**: 1-90 days (configurable via `retention-days`)
-- **Purpose**: Alternative access to memory data beyond cache expiration
-- **Use Case**: Long-term memory persistence for workflows that run infrequently
+- **Purpose**: Alternative access to cache data beyond cache expiration
+- **Use Case**: Long-term persistence for workflows that run infrequently
 
 #### Cache Scoping
 - **Branch Scoping**: Caches are accessible across branches in the same repository
@@ -151,20 +149,34 @@ custom-
 
 This ensures the most specific match is found first, with progressive fallbacks.
 
-## MCP Server Configuration
+## File Access and Organization
 
-The memory from the cache is accessed via a memory server is implicitly configured using npx following official MCP documentation:
+The cache folder provides standard file system access:
 
-```json
-"memory": {
-  "command": "npx",
-  "args": [
-    "@modelcontextprotocol/server-memory"
-  ],
-  "env": {
-    "MEMORY_FILE_PATH": "/tmp/cache-memory/memory.json"
-  }
-}
+### File Operations
+```
+/tmp/cache-memory/notes.txt           # Simple text files
+/tmp/cache-memory/config.json         # Structured data
+/tmp/cache-memory/logs/activity.log   # Organized in subdirectories
+/tmp/cache-memory/state/session.yaml  # State management
+```
+
+### Best Practices for File Organization
+
+Use descriptive file names and directory structures:
+
+```
+/tmp/cache-memory/
+├── preferences/
+│   ├── user-settings.json
+│   └── workflow-config.yaml
+├── logs/
+│   ├── activity.log
+│   └── errors.log
+├── state/
+│   ├── last-run.txt
+│   └── context.json
+└── notes.md
 ```
 
 ## Best Practices
@@ -181,9 +193,9 @@ tools:
 
 ### Memory Scope
 
-Consider the scope of memory needed:
+Consider the scope of cache needed:
 
-- **Workflow-specific**: Default behavior, memory per workflow
+- **Workflow-specific**: Default behavior, cache per workflow
 - **Repository-wide**: Use repository name in cache key
 - **User-specific**: Include user information in cache key
 
@@ -191,7 +203,7 @@ Consider the scope of memory needed:
 
 Be mindful of cache usage:
 
-- **Memory Size**: Monitor memory data growth over time
+- **File Size**: Monitor cache data growth over time
 - **Cache Limits**: Respect GitHub's 10GB repository cache limit
 - **Cleanup Strategy**: Consider periodic cache clearing for long-running projects
 
@@ -199,15 +211,15 @@ Be mindful of cache usage:
 
 ### Common Issues
 
-#### Memory Not Persisting
+#### Files Not Persisting
 - **Check Cache Keys**: Ensure keys are consistent between runs
-- **Verify Paths**: Confirm `/tmp/cache-memory` directory exists
+- **Verify Paths**: Confirm `/tmp/cache-memory/` directory exists
 - **Review Logs**: Check workflow logs for cache restore/save messages
 
-#### Package Installation Issues
-- **npm Registry Access**: Verify runner can access npm registry
-- **Package Availability**: Confirm `@modelcontextprotocol/server-memory` package exists
-- **Network Access**: Ensure runner has internet connectivity for package installation
+#### File Access Issues
+- **Directory Creation**: Ensure subdirectories are created before use
+- **Permissions**: Verify file write permissions in the cache folder
+- **Path Resolution**: Use absolute paths within `/tmp/cache-memory/`
 
 #### Cache Size Issues
 - **Monitor Usage**: Track cache size growth over time
@@ -230,29 +242,30 @@ timeout_minutes: 10  # Allow time for debugging
 
 Please debug the cache-memory functionality by:
 
-1. Checking what's currently in memory
-2. Storing a test message
-3. Retrieving the stored message
-4. Reporting on memory persistence
+1. Checking what files exist in the cache folder
+2. Creating a test file with current timestamp
+3. Reading the test file back
+4. Listing all files in the cache folder
+5. Reporting on file persistence
 ```
 
 ## Security Considerations
 
 ### Data Privacy
 
-- **Sensitive Data**: Avoid storing sensitive information in memory
-- **Access Control**: Memory data follows repository access permissions
+- **Sensitive Data**: Avoid storing sensitive information in cache files
+- **Access Control**: Cache data follows repository access permissions
 - **Audit Trail**: Cache access is logged in workflow execution logs
 
-### Package Security
+### File Security
 
-- **Official Package**: Use only the official `@modelcontextprotocol/server-memory` package
-- **Dependency Scanning**: npm automatically scans for vulnerabilities
-- **Audit Trail**: Package installation is logged in workflow execution logs
+- **Standard Permissions**: Files use standard GitHub Actions runner permissions
+- **Temporary Storage**: Cache directory is temporary and cleaned between runs
+- **No External Access**: Cache folder is only accessible within workflow execution
 
 ## Examples
 
-### Basic Memory Usage
+### Basic File Storage
 
 ```yaml
 ---
@@ -260,8 +273,8 @@ engine: claude
 on:
   workflow_dispatch:
     inputs:
-      remember:
-        description: 'Information to remember'
+      note:
+        description: 'Note to remember'
         required: true
 
 tools:
@@ -270,19 +283,19 @@ tools:
     allowed: [get_repository]
 ---
 
-# Memory Test Workflow
+# File Storage Test Workflow
 
-Store and retrieve information across workflow runs.
+Store and retrieve information using simple file operations.
 
 ## Task
 
-1. Check what you remember from previous runs
-2. Store the new information: "${{ inputs.remember }}"
-3. List all stored memories
-4. Provide a summary of memory persistence
+1. Check what files exist in the cache folder from previous runs
+2. Store the new note: "${{ inputs.note }}" in a timestamped file
+3. List all files in the cache folder
+4. Provide a summary of stored files and persistence
 ```
 
-### Project-Specific Memory
+### Project-Specific Cache
 
 ```yaml
 ---
@@ -296,27 +309,42 @@ tools:
 
 # Documentation Assistant
 
-Use project-specific memory to maintain context about documentation updates.
+Use project-specific cache to maintain context about documentation updates.
+Store progress, preferences, and notes in organized files.
 ```
 
-### Multi-Workflow Memory Sharing
+### Multi-Workflow File Sharing
 
 ```yaml
 ---
 engine: claude
 tools:
   cache-memory:
-    key: shared-memory-${{ github.repository }}
+    key: shared-cache-${{ github.repository }}
 ---
 
-# Shared Memory Workflow
+# Shared Cache Workflow
 
-Share memory data across multiple workflows in the same repository.
+Share cache data across multiple workflows in the same repository using files.
 ```
 
+## Migration from MCP Memory Server
+
+If migrating from the previous MCP memory server approach:
+
+### Changes
+- **No MCP server**: Cache-memory no longer uses `@modelcontextprotocol/server-memory`
+- **File operations**: Use standard file read/write instead of memory tools
+- **Direct access**: Access files directly at `/tmp/cache-memory/`
+- **No tools**: No `mcp__memory` tool is provided
+
+### Migration Steps
+1. **Update workflow syntax**: Remove any `docker-image` configuration
+2. **Modify prompts**: Update prompts to reference file operations instead of memory tools
+3. **Test file access**: Verify file operations work as expected
+
 ## Related Documentation
 
 - [Frontmatter Options](/gh-aw/reference/frontmatter/) - Complete frontmatter configuration guide
-- [MCP Tools](/gh-aw/guides/mcps/) - Model Context Protocol integration
 - [Safe Outputs](/gh-aw/reference/safe-outputs/) - Output processing and automation
 - [GitHub Actions Cache Documentation](https://docs.github.com/en/actions/using-workflows/caching-dependencies-to-speed-up-workflows) - Official GitHub cache documentation
\ No newline at end of file
diff --git a/pkg/cli/workflows/test-claude-cache-memory.lock.yml b/pkg/cli/workflows/test-claude-cache-memory.lock.yml
index bcf5c92e59..0529c8fcfe 100644
--- a/pkg/cli/workflows/test-claude-cache-memory.lock.yml
+++ b/pkg/cli/workflows/test-claude-cache-memory.lock.yml
@@ -2,7 +2,7 @@
 # To update this file, edit the corresponding .md file and run:
 #   gh aw compile
 
-name: "Test Claude with Cache Memory and Custom Docker Image"
+name: "Test Claude with Cache Memory File Share"
 on:
   workflow_dispatch:
     inputs:
@@ -16,19 +16,23 @@ permissions: {}
 concurrency:
   group: "gh-aw-${{ github.workflow }}"
 
-run-name: "Test Claude with Cache Memory and Custom Docker Image"
+run-name: "Test Claude with Cache Memory File Share"
 
 jobs:
-  test-claude-with-cache-memory-and-custom-docker-image:
+  test-claude-with-cache-memory-file-share:
     runs-on: ubuntu-latest
     permissions: read-all
     steps:
       - name: Checkout repository
         uses: actions/checkout@v5
-      # Cache memory MCP configuration from frontmatter processed below
+      # Cache memory file share configuration from frontmatter processed below
       - name: Create cache-memory directory
-        run: mkdir -p /tmp/cache-memory
-      - name: Cache memory MCP data
+        run: |
+          mkdir -p /tmp/cache-memory
+          echo "Cache memory directory created at /tmp/cache-memory"
+          echo "This folder provides persistent file storage across workflow runs"
+          echo "LLMs and agentic tools can freely read and write files in this directory"
+      - name: Cache memory file share data
         uses: actions/cache@v4
         with:
           key: memory-${{ github.workflow }}-${{ github.run_id }}
@@ -36,10 +40,10 @@ jobs:
           restore-keys: |
             memory-${{ github.workflow }}-
             memory-
-      - name: Upload memory MCP data as artifact
+      - name: Upload cache-memory data as artifact
         uses: actions/upload-artifact@v4
         with:
-          name: cache-memory-data
+          name: cache-memory
           path: /tmp/cache-memory
           retention-days: 14
       - name: Generate Claude Settings
@@ -156,15 +160,6 @@ jobs:
           cat > /tmp/mcp-config/mcp-servers.json << 'EOF'
           {
             "mcpServers": {
-              "memory": {
-                "command": "npx",
-                "args": [
-                  "@modelcontextprotocol/server-memory"
-                ],
-                "env": {
-                  "MEMORY_FILE_PATH": "/tmp/cache-memory/memory.json"
-                }
-              },
               "github": {
                 "command": "docker",
                 "args": [
@@ -188,45 +183,64 @@ jobs:
         run: |
           mkdir -p /tmp/aw-prompts
           cat > $GITHUB_AW_PROMPT << 'EOF'
-          # Test Claude with Cache Memory and Custom Docker Image
+          # Test Claude with Cache Memory File Share
           
-          You are a test agent that demonstrates the cache-memory functionality with Claude engine using a custom Docker image.
+          You are a test agent that demonstrates the cache-memory functionality with Claude engine using a simple file share approach.
           
           ## Task
           
           Your job is to:
           
-          1. **Store a test task** in your memory using the memory MCP server
-          2. **Retrieve any previous tasks** that you've stored in memory
-          3. **Report on the memory contents** including both current and historical tasks
+          1. **Store a test task** in the cache folder using file operations
+          2. **Retrieve any previous tasks** that you've stored in previous runs
+          3. **Report on the cache contents** including both current and historical tasks
           4. **Use GitHub tools** to get basic repository information
           
           ## Instructions
           
-          1. First, use the memory tool to see what you already know from previous runs
-          2. Store a new test task: "Test task for run ${{ github.run_number }}" in your memory
-          3. List all tasks you now have in memory
+          1. First, check what files exist in `/tmp/cache-memory/` from previous runs
+          2. Store a new test task: "Test task for run ${{ github.run_number }}" in a file in the cache folder
+          3. List all files and contents you now have in the cache folder
           4. Get basic information about this repository using the GitHub tool
           5. Provide a summary of:
-             - What you remembered from before
+             - What you found from before (if anything)
              - What you just stored
              - Basic repository information
           
           ## Expected Behavior
           
-          - **First run**: Should show empty memory, then store the new task
-          - **Subsequent runs**: Should show previously stored tasks, then add the new one
-          - **Memory persistence**: Tasks should persist across workflow runs thanks to cache-memory
-          - **Custom Docker image**: Uses ghcr.io/modelcontextprotocol/server-memory:v1.0.0 instead of latest
-          - **Artifact upload**: Memory data is also uploaded as artifact with 14-day retention
+          - **First run**: Should show empty cache folder, then store the new task
+          - **Subsequent runs**: Should show previously stored files, then add the new one
+          - **File persistence**: Files should persist across workflow runs thanks to cache-memory
+          - **Simple file access**: Uses standard file operations (no MCP server needed)
+          - **Artifact upload**: Cache data is also uploaded as artifact with 14-day retention
           
           This workflow tests that the cache-memory configuration properly:
-          - Mounts the memory MCP server with custom Docker image
+          - Creates a simple file share at `/tmp/cache-memory/`
           - Persists data between runs using GitHub Actions cache
-          - Uploads memory data as artifacts with configurable retention
-          - Works with Claude engine and MCP tools
+          - Uploads cache data as artifacts with configurable retention
+          - Works with Claude engine and file operations
           - Integrates with other tools like GitHub
           
+          
+          ---
+          
+          ## Cache Folder Available
+          
+          You have access to a persistent cache folder at `/tmp/cache-memory/` where you can read and write files to create memories and store information.
+          
+          - **Read/Write Access**: You can freely read from and write to any files in this folder
+          - **Persistence**: Files in this folder persist across workflow runs via GitHub Actions cache
+          - **Last Write Wins**: If multiple processes write to the same file, the last write will be preserved
+          - **File Share**: Use this as a simple file share - organize files as you see fit
+          
+          Examples of what you can store:
+          - `/tmp/cache-memory/notes.txt` - general notes and observations
+          - `/tmp/cache-memory/preferences.json` - user preferences and settings
+          - `/tmp/cache-memory/history.log` - activity history and logs
+          - `/tmp/cache-memory/state/` - organized state files in subdirectories
+          
+          Feel free to create, read, update, and organize files in this folder as needed for your tasks.
           EOF
       - name: Print prompt to step summary
         run: |
@@ -248,7 +262,7 @@ jobs:
               engine_name: "Claude Code",
               model: "",
               version: "",
-              workflow_name: "Test Claude with Cache Memory and Custom Docker Image",
+              workflow_name: "Test Claude with Cache Memory File Share",
               experimental: false,
               supports_tools_whitelist: true,
               supports_http_transport: true,
@@ -279,14 +293,18 @@ jobs:
       - name: Execute Claude Code CLI
         id: agentic_execution
         # Allowed tools (sorted):
+        # - Edit(/tmp/cache-memory/*)
         # - ExitPlanMode
         # - Glob
         # - Grep
         # - LS
+        # - MultiEdit(/tmp/cache-memory/*)
         # - NotebookRead
         # - Read
+        # - Read(/tmp/cache-memory/*)
         # - Task
         # - TodoWrite
+        # - Write(/tmp/cache-memory/*)
         # - mcp__github__download_workflow_run_artifact
         # - mcp__github__get_code_scanning_alert
         # - mcp__github__get_commit
@@ -332,12 +350,11 @@ jobs:
         # - mcp__github__search_pull_requests
         # - mcp__github__search_repositories
         # - mcp__github__search_users
-        # - mcp__memory
         timeout-minutes: 5
         run: |
           set -o pipefail
           # Execute Claude Code CLI with prompt from file
-          npx @anthropic-ai/claude-code@latest --print --mcp-config /tmp/mcp-config/mcp-servers.json --allowed-tools "ExitPlanMode,Glob,Grep,LS,NotebookRead,Read,Task,TodoWrite,mcp__github__download_workflow_run_artifact,mcp__github__get_code_scanning_alert,mcp__github__get_commit,mcp__github__get_dependabot_alert,mcp__github__get_discussion,mcp__github__get_discussion_comments,mcp__github__get_file_contents,mcp__github__get_issue,mcp__github__get_issue_comments,mcp__github__get_job_logs,mcp__github__get_me,mcp__github__get_notification_details,mcp__github__get_pull_request,mcp__github__get_pull_request_comments,mcp__github__get_pull_request_diff,mcp__github__get_pull_request_files,mcp__github__get_pull_request_reviews,mcp__github__get_pull_request_status,mcp__github__get_repository,mcp__github__get_secret_scanning_alert,mcp__github__get_tag,mcp__github__get_workflow_run,mcp__github__get_workflow_run_logs,mcp__github__get_workflow_run_usage,mcp__github__list_branches,mcp__github__list_code_scanning_alerts,mcp__github__list_commits,mcp__github__list_dependabot_alerts,mcp__github__list_discussion_categories,mcp__github__list_discussions,mcp__github__list_issues,mcp__github__list_notifications,mcp__github__list_pull_requests,mcp__github__list_secret_scanning_alerts,mcp__github__list_tags,mcp__github__list_workflow_jobs,mcp__github__list_workflow_run_artifacts,mcp__github__list_workflow_runs,mcp__github__list_workflows,mcp__github__search_code,mcp__github__search_issues,mcp__github__search_orgs,mcp__github__search_pull_requests,mcp__github__search_repositories,mcp__github__search_users,mcp__memory" --debug --verbose --permission-mode bypassPermissions --output-format json --settings /tmp/.claude/settings.json "$(cat /tmp/aw-prompts/prompt.txt)" 2>&1 | tee /tmp/test-claude-with-cache-memory-and-custom-docker-image.log
+          npx @anthropic-ai/claude-code@latest --print --mcp-config /tmp/mcp-config/mcp-servers.json --allowed-tools "Edit(/tmp/cache-memory/*),ExitPlanMode,Glob,Grep,LS,MultiEdit(/tmp/cache-memory/*),NotebookRead,Read,Read(/tmp/cache-memory/*),Task,TodoWrite,Write(/tmp/cache-memory/*),mcp__github__download_workflow_run_artifact,mcp__github__get_code_scanning_alert,mcp__github__get_commit,mcp__github__get_dependabot_alert,mcp__github__get_discussion,mcp__github__get_discussion_comments,mcp__github__get_file_contents,mcp__github__get_issue,mcp__github__get_issue_comments,mcp__github__get_job_logs,mcp__github__get_me,mcp__github__get_notification_details,mcp__github__get_pull_request,mcp__github__get_pull_request_comments,mcp__github__get_pull_request_diff,mcp__github__get_pull_request_files,mcp__github__get_pull_request_reviews,mcp__github__get_pull_request_status,mcp__github__get_repository,mcp__github__get_secret_scanning_alert,mcp__github__get_tag,mcp__github__get_workflow_run,mcp__github__get_workflow_run_logs,mcp__github__get_workflow_run_usage,mcp__github__list_branches,mcp__github__list_code_scanning_alerts,mcp__github__list_commits,mcp__github__list_dependabot_alerts,mcp__github__list_discussion_categories,mcp__github__list_discussions,mcp__github__list_issues,mcp__github__list_notifications,mcp__github__list_pull_requests,mcp__github__list_secret_scanning_alerts,mcp__github__list_tags,mcp__github__list_workflow_jobs,mcp__github__list_workflow_run_artifacts,mcp__github__list_workflow_runs,mcp__github__list_workflows,mcp__github__search_code,mcp__github__search_issues,mcp__github__search_orgs,mcp__github__search_pull_requests,mcp__github__search_repositories,mcp__github__search_users" --debug --verbose --permission-mode bypassPermissions --output-format json --settings /tmp/.claude/settings.json "$(cat /tmp/aw-prompts/prompt.txt)" 2>&1 | tee /tmp/test-claude-with-cache-memory-file-share.log
         env:
           ANTHROPIC_API_KEY: ${{ secrets.ANTHROPIC_API_KEY }}
           DISABLE_TELEMETRY: "1"
@@ -348,15 +365,15 @@ jobs:
         if: always()
         run: |
           # Ensure log file exists
-          touch /tmp/test-claude-with-cache-memory-and-custom-docker-image.log
+          touch /tmp/test-claude-with-cache-memory-file-share.log
           # Show last few lines for debugging
           echo "=== Last 10 lines of Claude execution log ==="
-          tail -10 /tmp/test-claude-with-cache-memory-and-custom-docker-image.log || echo "No log content available"
+          tail -10 /tmp/test-claude-with-cache-memory-file-share.log || echo "No log content available"
       - name: Parse agent logs for step summary
         if: always()
         uses: actions/github-script@v7
         env:
-          GITHUB_AW_AGENT_OUTPUT: /tmp/test-claude-with-cache-memory-and-custom-docker-image.log
+          GITHUB_AW_AGENT_OUTPUT: /tmp/test-claude-with-cache-memory-file-share.log
         with:
           script: |
             function main() {
@@ -873,7 +890,7 @@ jobs:
         if: always()
         uses: actions/upload-artifact@v4
         with:
-          name: test-claude-with-cache-memory-and-custom-docker-image.log
-          path: /tmp/test-claude-with-cache-memory-and-custom-docker-image.log
+          name: test-claude-with-cache-memory-file-share.log
+          path: /tmp/test-claude-with-cache-memory-file-share.log
           if-no-files-found: warn
 
diff --git a/pkg/cli/workflows/test-claude-cache-memory.md b/pkg/cli/workflows/test-claude-cache-memory.md
index 75ce252002..c937df10b2 100644
--- a/pkg/cli/workflows/test-claude-cache-memory.md
+++ b/pkg/cli/workflows/test-claude-cache-memory.md
@@ -10,7 +10,6 @@ on:
 
 tools:
   cache-memory:
-    docker-image: "ghcr.io/modelcontextprotocol/server-memory:v1.0.0"
     retention-days: 14
   github:
     allowed: [get_repository]
@@ -18,41 +17,41 @@ tools:
 timeout_minutes: 5
 ---
 
-# Test Claude with Cache Memory and Custom Docker Image
+# Test Claude with Cache Memory File Share
 
-You are a test agent that demonstrates the cache-memory functionality with Claude engine using a custom Docker image.
+You are a test agent that demonstrates the cache-memory functionality with Claude engine using a simple file share approach.
 
 ## Task
 
 Your job is to:
 
-1. **Store a test task** in your memory using the memory MCP server
-2. **Retrieve any previous tasks** that you've stored in memory
-3. **Report on the memory contents** including both current and historical tasks
+1. **Store a test task** in the cache folder using file operations
+2. **Retrieve any previous tasks** that you've stored in previous runs
+3. **Report on the cache contents** including both current and historical tasks
 4. **Use GitHub tools** to get basic repository information
 
 ## Instructions
 
-1. First, use the memory tool to see what you already know from previous runs
-2. Store a new test task: "Test task for run ${{ github.run_number }}" in your memory
-3. List all tasks you now have in memory
+1. First, check what files exist in `/tmp/cache-memory/` from previous runs
+2. Store a new test task: "Test task for run ${{ github.run_number }}" in a file in the cache folder
+3. List all files and contents you now have in the cache folder
 4. Get basic information about this repository using the GitHub tool
 5. Provide a summary of:
-   - What you remembered from before
+   - What you found from before (if anything)
    - What you just stored
    - Basic repository information
 
 ## Expected Behavior
 
-- **First run**: Should show empty memory, then store the new task
-- **Subsequent runs**: Should show previously stored tasks, then add the new one
-- **Memory persistence**: Tasks should persist across workflow runs thanks to cache-memory
-- **Custom Docker image**: Uses ghcr.io/modelcontextprotocol/server-memory:v1.0.0 instead of latest
-- **Artifact upload**: Memory data is also uploaded as artifact with 14-day retention
+- **First run**: Should show empty cache folder, then store the new task
+- **Subsequent runs**: Should show previously stored files, then add the new one
+- **File persistence**: Files should persist across workflow runs thanks to cache-memory
+- **Simple file access**: Uses standard file operations (no MCP server needed)
+- **Artifact upload**: Cache data is also uploaded as artifact with 14-day retention
 
 This workflow tests that the cache-memory configuration properly:
-- Mounts the memory MCP server with custom Docker image
+- Creates a simple file share at `/tmp/cache-memory/`
 - Persists data between runs using GitHub Actions cache
-- Uploads memory data as artifacts with configurable retention
-- Works with Claude engine and MCP tools
+- Uploads cache data as artifacts with configurable retention
+- Works with Claude engine and file operations
 - Integrates with other tools like GitHub
\ No newline at end of file
diff --git a/pkg/cli/workflows/test-claude-missing-tool.lock.yml b/pkg/cli/workflows/test-claude-missing-tool.lock.yml
index 92d43c84d1..9d5cadbd57 100644
--- a/pkg/cli/workflows/test-claude-missing-tool.lock.yml
+++ b/pkg/cli/workflows/test-claude-missing-tool.lock.yml
@@ -27,10 +27,14 @@ jobs:
     steps:
       - name: Checkout repository
         uses: actions/checkout@v5
-      # Cache memory MCP configuration from frontmatter processed below
+      # Cache memory file share configuration from frontmatter processed below
       - name: Create cache-memory directory
-        run: mkdir -p /tmp/cache-memory
-      - name: Cache memory MCP data
+        run: |
+          mkdir -p /tmp/cache-memory
+          echo "Cache memory directory created at /tmp/cache-memory"
+          echo "This folder provides persistent file storage across workflow runs"
+          echo "LLMs and agentic tools can freely read and write files in this directory"
+      - name: Cache memory file share data
         uses: actions/cache@v4
         with:
           key: memory-${{ github.workflow }}-${{ github.run_id }}
@@ -38,6 +42,11 @@ jobs:
           restore-keys: |
             memory-${{ github.workflow }}-
             memory-
+      - name: Upload cache-memory data as artifact
+        uses: actions/upload-artifact@v4
+        with:
+          name: cache-memory
+          path: /tmp/cache-memory
       - name: Generate Claude Settings
         run: |
           mkdir -p /tmp/.claude
@@ -610,15 +619,6 @@ jobs:
           cat > /tmp/mcp-config/mcp-servers.json << 'EOF'
           {
             "mcpServers": {
-              "memory": {
-                "command": "npx",
-                "args": [
-                  "@modelcontextprotocol/server-memory"
-                ],
-                "env": {
-                  "MEMORY_FILE_PATH": "/tmp/cache-memory/memory.json"
-                }
-              },
               "github": {
                 "command": "docker",
                 "args": [
@@ -691,6 +691,25 @@ jobs:
           - Safe output system handles missing tool scenarios correctly
           
           
+          ---
+          
+          ## Cache Folder Available
+          
+          You have access to a persistent cache folder at `/tmp/cache-memory/` where you can read and write files to create memories and store information.
+          
+          - **Read/Write Access**: You can freely read from and write to any files in this folder
+          - **Persistence**: Files in this folder persist across workflow runs via GitHub Actions cache
+          - **Last Write Wins**: If multiple processes write to the same file, the last write will be preserved
+          - **File Share**: Use this as a simple file share - organize files as you see fit
+          
+          Examples of what you can store:
+          - `/tmp/cache-memory/notes.txt` - general notes and observations
+          - `/tmp/cache-memory/preferences.json` - user preferences and settings
+          - `/tmp/cache-memory/history.log` - activity history and logs
+          - `/tmp/cache-memory/state/` - organized state files in subdirectories
+          
+          Feel free to create, read, update, and organize files in this folder as needed for your tasks.
+          
           ---
           
           ## Reporting Missing Tools or Functionality
@@ -748,15 +767,19 @@ jobs:
       - name: Execute Claude Code CLI
         id: agentic_execution
         # Allowed tools (sorted):
+        # - Edit(/tmp/cache-memory/*)
         # - ExitPlanMode
         # - Glob
         # - Grep
         # - LS
+        # - MultiEdit(/tmp/cache-memory/*)
         # - NotebookRead
         # - Read
+        # - Read(/tmp/cache-memory/*)
         # - Task
         # - TodoWrite
         # - Write
+        # - Write(/tmp/cache-memory/*)
         # - mcp__github__download_workflow_run_artifact
         # - mcp__github__get_code_scanning_alert
         # - mcp__github__get_commit
@@ -802,12 +825,11 @@ jobs:
         # - mcp__github__search_pull_requests
         # - mcp__github__search_repositories
         # - mcp__github__search_users
-        # - mcp__memory
         timeout-minutes: 5
         run: |
           set -o pipefail
           # Execute Claude Code CLI with prompt from file
-          npx @anthropic-ai/claude-code@latest --print --mcp-config /tmp/mcp-config/mcp-servers.json --allowed-tools "ExitPlanMode,Glob,Grep,LS,NotebookRead,Read,Task,TodoWrite,Write,mcp__github__download_workflow_run_artifact,mcp__github__get_code_scanning_alert,mcp__github__get_commit,mcp__github__get_dependabot_alert,mcp__github__get_discussion,mcp__github__get_discussion_comments,mcp__github__get_file_contents,mcp__github__get_issue,mcp__github__get_issue_comments,mcp__github__get_job_logs,mcp__github__get_me,mcp__github__get_notification_details,mcp__github__get_pull_request,mcp__github__get_pull_request_comments,mcp__github__get_pull_request_diff,mcp__github__get_pull_request_files,mcp__github__get_pull_request_reviews,mcp__github__get_pull_request_status,mcp__github__get_repository,mcp__github__get_secret_scanning_alert,mcp__github__get_tag,mcp__github__get_workflow_run,mcp__github__get_workflow_run_logs,mcp__github__get_workflow_run_usage,mcp__github__list_branches,mcp__github__list_code_scanning_alerts,mcp__github__list_commits,mcp__github__list_dependabot_alerts,mcp__github__list_discussion_categories,mcp__github__list_discussions,mcp__github__list_issues,mcp__github__list_notifications,mcp__github__list_pull_requests,mcp__github__list_secret_scanning_alerts,mcp__github__list_tags,mcp__github__list_workflow_jobs,mcp__github__list_workflow_run_artifacts,mcp__github__list_workflow_runs,mcp__github__list_workflows,mcp__github__search_code,mcp__github__search_issues,mcp__github__search_orgs,mcp__github__search_pull_requests,mcp__github__search_repositories,mcp__github__search_users,mcp__memory" --debug --verbose --permission-mode bypassPermissions --output-format json --settings /tmp/.claude/settings.json "$(cat /tmp/aw-prompts/prompt.txt)" 2>&1 | tee /tmp/test-claude-with-missing-tool-safe-output-and-cache-memory.log
+          npx @anthropic-ai/claude-code@latest --print --mcp-config /tmp/mcp-config/mcp-servers.json --allowed-tools "Edit(/tmp/cache-memory/*),ExitPlanMode,Glob,Grep,LS,MultiEdit(/tmp/cache-memory/*),NotebookRead,Read,Read(/tmp/cache-memory/*),Task,TodoWrite,Write,Write(/tmp/cache-memory/*),mcp__github__download_workflow_run_artifact,mcp__github__get_code_scanning_alert,mcp__github__get_commit,mcp__github__get_dependabot_alert,mcp__github__get_discussion,mcp__github__get_discussion_comments,mcp__github__get_file_contents,mcp__github__get_issue,mcp__github__get_issue_comments,mcp__github__get_job_logs,mcp__github__get_me,mcp__github__get_notification_details,mcp__github__get_pull_request,mcp__github__get_pull_request_comments,mcp__github__get_pull_request_diff,mcp__github__get_pull_request_files,mcp__github__get_pull_request_reviews,mcp__github__get_pull_request_status,mcp__github__get_repository,mcp__github__get_secret_scanning_alert,mcp__github__get_tag,mcp__github__get_workflow_run,mcp__github__get_workflow_run_logs,mcp__github__get_workflow_run_usage,mcp__github__list_branches,mcp__github__list_code_scanning_alerts,mcp__github__list_commits,mcp__github__list_dependabot_alerts,mcp__github__list_discussion_categories,mcp__github__list_discussions,mcp__github__list_issues,mcp__github__list_notifications,mcp__github__list_pull_requests,mcp__github__list_secret_scanning_alerts,mcp__github__list_tags,mcp__github__list_workflow_jobs,mcp__github__list_workflow_run_artifacts,mcp__github__list_workflow_runs,mcp__github__list_workflows,mcp__github__search_code,mcp__github__search_issues,mcp__github__search_orgs,mcp__github__search_pull_requests,mcp__github__search_repositories,mcp__github__search_users" --debug --verbose --permission-mode bypassPermissions --output-format json --settings /tmp/.claude/settings.json "$(cat /tmp/aw-prompts/prompt.txt)" 2>&1 | tee /tmp/test-claude-with-missing-tool-safe-output-and-cache-memory.log
         env:
           ANTHROPIC_API_KEY: ${{ secrets.ANTHROPIC_API_KEY }}
           DISABLE_TELEMETRY: "1"
diff --git a/pkg/console/console.go b/pkg/console/console.go
index 1fa146f3fe..14816e6bce 100644
--- a/pkg/console/console.go
+++ b/pkg/console/console.go
@@ -225,8 +225,8 @@ func FormatWarningMessage(message string) string {
 var (
 	tableHeaderStyle = lipgloss.NewStyle().
 				Bold(true).
-				Foreground(lipgloss.Color("#BD93F9")).
-				Background(lipgloss.Color("#44475A"))
+				Foreground(lipgloss.Color("#FFFFFF")).
+				Background(lipgloss.Color("#34374A"))
 
 	tableCellStyle = lipgloss.NewStyle().
 			Foreground(lipgloss.Color("#F8F8F2"))
@@ -259,8 +259,7 @@ func RenderTable(config TableConfig) string {
 	if config.Title != "" {
 		titleStyle := lipgloss.NewStyle().
 			Bold(true).
-			Foreground(lipgloss.Color("#50FA7B")).
-			MarginBottom(1)
+			Foreground(lipgloss.Color("#50FA7B"))
 		output.WriteString(applyStyle(titleStyle, config.Title))
 		output.WriteString("\n")
 	}
diff --git a/pkg/workflow/cache.go b/pkg/workflow/cache.go
index 2c9bc20320..dc0d15d313 100644
--- a/pkg/workflow/cache.go
+++ b/pkg/workflow/cache.go
@@ -105,17 +105,22 @@ func generateCacheSteps(builder *strings.Builder, data *WorkflowData, verbose bo
 }
 
 // generateCacheMemorySteps generates cache steps for the cache-memory configuration
+// Cache-memory provides a simple file share that LLMs can read/write freely
 func generateCacheMemorySteps(builder *strings.Builder, data *WorkflowData, verbose bool) {
 	if data.CacheMemoryConfig == nil || !data.CacheMemoryConfig.Enabled {
 		return
 	}
 
 	// Add comment indicating cache-memory configuration was processed
-	builder.WriteString("      # Cache memory MCP configuration from frontmatter processed below\n")
+	builder.WriteString("      # Cache memory file share configuration from frontmatter processed below\n")
 
 	// Add step to create cache-memory directory
 	builder.WriteString("      - name: Create cache-memory directory\n")
-	builder.WriteString("        run: mkdir -p /tmp/cache-memory\n")
+	builder.WriteString("        run: |\n")
+	builder.WriteString("          mkdir -p /tmp/cache-memory\n")
+	builder.WriteString("          echo \"Cache memory directory created at /tmp/cache-memory\"\n")
+	builder.WriteString("          echo \"This folder provides persistent file storage across workflow runs\"\n")
+	builder.WriteString("          echo \"LLMs and agentic tools can freely read and write files in this directory\"\n")
 
 	// Use the parsed configuration
 	cacheKey := data.CacheMemoryConfig.Key
@@ -138,7 +143,7 @@ func generateCacheMemorySteps(builder *strings.Builder, data *WorkflowData, verb
 		restoreKeys = append(restoreKeys, restoreKey)
 	}
 
-	builder.WriteString("      - name: Cache memory MCP data\n")
+	builder.WriteString("      - name: Cache memory file share data\n")
 	builder.WriteString("        uses: actions/cache@v4\n")
 	builder.WriteString("        with:\n")
 	fmt.Fprintf(builder, "          key: %s\n", cacheKey)
@@ -148,13 +153,14 @@ func generateCacheMemorySteps(builder *strings.Builder, data *WorkflowData, verb
 		fmt.Fprintf(builder, "            %s\n", key)
 	}
 
-	// Add upload-artifact step if retention-days is configured
+	// Always add upload-artifact step for cache-memory (runs always)
+	builder.WriteString("      - name: Upload cache-memory data as artifact\n")
+	builder.WriteString("        uses: actions/upload-artifact@v4\n")
+	builder.WriteString("        with:\n")
+	builder.WriteString("          name: cache-memory\n")
+	builder.WriteString("          path: /tmp/cache-memory\n")
+	// Add retention-days if configured
 	if data.CacheMemoryConfig.RetentionDays != nil {
-		builder.WriteString("      - name: Upload memory MCP data as artifact\n")
-		builder.WriteString("        uses: actions/upload-artifact@v4\n")
-		builder.WriteString("        with:\n")
-		builder.WriteString("          name: cache-memory-data\n")
-		builder.WriteString("          path: /tmp/cache-memory\n")
 		fmt.Fprintf(builder, "          retention-days: %d\n", *data.CacheMemoryConfig.RetentionDays)
 	}
 }
diff --git a/pkg/workflow/claude_engine.go b/pkg/workflow/claude_engine.go
index 577f89dbf5..9356c8d6d5 100644
--- a/pkg/workflow/claude_engine.go
+++ b/pkg/workflow/claude_engine.go
@@ -448,11 +448,25 @@ func (e *ClaudeEngine) computeAllowedClaudeToolsString(tools map[string]any, saf
 			// Skip the claude section as we've already processed it
 			continue
 		} else {
-			// Handle cache-memory as a special case first (can be boolean or map)
+			// Handle cache-memory as a special case - it provides file system access but no MCP tool
 			if toolName == "cache-memory" {
-				// For cache-memory, it's configured as MCP server "memory" and has no allowed restrictions
-				// Default to wildcard access since cache-memory doesn't specify allowed tools
-				allowedTools = append(allowedTools, "mcp__memory")
+				// Cache-memory now provides simple file share access at /tmp/cache-memory/
+				// Add path-specific Read and Write tools for the cache directory only
+				cacheDirPattern := "/tmp/cache-memory/*"
+
+				// Add path-specific tools for cache directory access
+				if !slices.Contains(allowedTools, fmt.Sprintf("Read(%s)", cacheDirPattern)) {
+					allowedTools = append(allowedTools, fmt.Sprintf("Read(%s)", cacheDirPattern))
+				}
+				if !slices.Contains(allowedTools, fmt.Sprintf("Write(%s)", cacheDirPattern)) {
+					allowedTools = append(allowedTools, fmt.Sprintf("Write(%s)", cacheDirPattern))
+				}
+				if !slices.Contains(allowedTools, fmt.Sprintf("Edit(%s)", cacheDirPattern)) {
+					allowedTools = append(allowedTools, fmt.Sprintf("Edit(%s)", cacheDirPattern))
+				}
+				if !slices.Contains(allowedTools, fmt.Sprintf("MultiEdit(%s)", cacheDirPattern)) {
+					allowedTools = append(allowedTools, fmt.Sprintf("MultiEdit(%s)", cacheDirPattern))
+				}
 				continue
 			}
 
@@ -652,23 +666,13 @@ func (e *ClaudeEngine) renderClaudeMCPConfig(yaml *strings.Builder, toolName str
 	return nil
 }
 
-// renderCacheMemoryMCPConfig generates the Memory MCP server configuration
-// Uses npx-based @modelcontextprotocol/server-memory setup
+// renderCacheMemoryMCPConfig handles cache-memory configuration without MCP server mounting
+// Cache-memory is now a simple file share, not an MCP server
 func (e *ClaudeEngine) renderCacheMemoryMCPConfig(yaml *strings.Builder, isLast bool, workflowData *WorkflowData) {
-	yaml.WriteString("              \"memory\": {\n")
-	yaml.WriteString("                \"command\": \"npx\",\n")
-	yaml.WriteString("                \"args\": [\n")
-	yaml.WriteString("                  \"@modelcontextprotocol/server-memory\"\n")
-	yaml.WriteString("                ],\n")
-	yaml.WriteString("                \"env\": {\n")
-	yaml.WriteString("                  \"MEMORY_FILE_PATH\": \"/tmp/cache-memory/memory.json\"\n")
-	yaml.WriteString("                }\n")
-
-	if isLast {
-		yaml.WriteString("              }\n")
-	} else {
-		yaml.WriteString("              },\n")
-	}
+	// Cache-memory no longer uses MCP server mounting
+	// The cache folder is available as a simple file share at /tmp/cache-memory/
+	// The folder is created by the cache step and is accessible to all tools
+	// No MCP configuration is needed for simple file access
 }
 
 // renderSafeOutputsMCPConfig generates the Safe Outputs MCP server configuration
diff --git a/pkg/workflow/claude_engine_tools_test.go b/pkg/workflow/claude_engine_tools_test.go
index de422665c7..35f00adf1f 100644
--- a/pkg/workflow/claude_engine_tools_test.go
+++ b/pkg/workflow/claude_engine_tools_test.go
@@ -50,20 +50,20 @@ func TestClaudeEngineComputeAllowedTools(t *testing.T) {
 			expected: "ExitPlanMode,Glob,Grep,LS,NotebookRead,Read,Task,TodoWrite,mcp__github__create_issue,mcp__github__list_issues",
 		},
 		{
-			name: "cache-memory tool (should get wildcard access as mcp__memory)",
+			name: "cache-memory tool (provides file system access with path-specific cache tools)",
 			tools: map[string]any{
 				"cache-memory": map[string]any{
 					"key": "test-memory-key",
 				},
 			},
-			expected: "ExitPlanMode,Glob,Grep,LS,NotebookRead,Read,Task,TodoWrite,mcp__memory",
+			expected: "Edit(/tmp/cache-memory/*),ExitPlanMode,Glob,Grep,LS,MultiEdit(/tmp/cache-memory/*),NotebookRead,Read,Read(/tmp/cache-memory/*),Task,TodoWrite,Write(/tmp/cache-memory/*)",
 		},
 		{
 			name: "cache-memory with boolean true",
 			tools: map[string]any{
 				"cache-memory": true,
 			},
-			expected: "ExitPlanMode,Glob,Grep,LS,NotebookRead,Read,Task,TodoWrite,mcp__memory",
+			expected: "Edit(/tmp/cache-memory/*),ExitPlanMode,Glob,Grep,LS,MultiEdit(/tmp/cache-memory/*),NotebookRead,Read,Read(/tmp/cache-memory/*),Task,TodoWrite,Write(/tmp/cache-memory/*)",
 		},
 		{
 			name: "cache-memory with github tools",
@@ -73,7 +73,7 @@ func TestClaudeEngineComputeAllowedTools(t *testing.T) {
 					"allowed": []any{"get_repository"},
 				},
 			},
-			expected: "ExitPlanMode,Glob,Grep,LS,NotebookRead,Read,Task,TodoWrite,mcp__github__get_repository,mcp__memory",
+			expected: "Edit(/tmp/cache-memory/*),ExitPlanMode,Glob,Grep,LS,MultiEdit(/tmp/cache-memory/*),NotebookRead,Read,Read(/tmp/cache-memory/*),Task,TodoWrite,Write(/tmp/cache-memory/*),mcp__github__get_repository",
 		},
 		{
 			name: "mixed neutral and mcp tools",
diff --git a/pkg/workflow/compiler.go b/pkg/workflow/compiler.go
index 62d28b23e8..5149d99cbc 100644
--- a/pkg/workflow/compiler.go
+++ b/pkg/workflow/compiler.go
@@ -156,7 +156,6 @@ type WorkflowData struct {
 type CacheMemoryConfig struct {
 	Enabled       bool   `yaml:"enabled,omitempty"`        // whether cache-memory is enabled
 	Key           string `yaml:"key,omitempty"`            // custom cache key
-	DockerImage   string `yaml:"docker-image,omitempty"`   // deprecated: no longer used (npx is used instead)
 	RetentionDays *int   `yaml:"retention-days,omitempty"` // retention days for upload-artifact action
 }
 
@@ -626,6 +625,12 @@ func (c *Compiler) parseWorkflowFile(markdownPath string) (*WorkflowData, error)
 		return nil, fmt.Errorf("failed to extract workflow name: %w", err)
 	}
 
+	// Check if frontmatter specifies a custom name and use it instead
+	frontmatterName := c.extractStringValue(result.Frontmatter, "name")
+	if frontmatterName != "" {
+		workflowName = frontmatterName
+	}
+
 	if c.verbose {
 		fmt.Println(console.FormatInfoMessage(fmt.Sprintf("Extracted workflow name: '%s'", workflowName)))
 	}
@@ -636,7 +641,7 @@ func (c *Compiler) parseWorkflowFile(markdownPath string) (*WorkflowData, error)
 	// Build workflow data
 	workflowData := &WorkflowData{
 		Name:               workflowName,
-		FrontmatterName:    c.extractStringValue(result.Frontmatter, "name"),
+		FrontmatterName:    frontmatterName,
 		Tools:              tools,
 		MarkdownContent:    markdownContent,
 		AI:                 engineSetting,
@@ -3528,6 +3533,29 @@ func (c *Compiler) generatePrompt(yaml *strings.Builder, data *WorkflowData) {
 		yaml.WriteString("          " + line + "\n")
 	}
 
+	// Add cache folder notification if cache-memory is enabled
+	if data.CacheMemoryConfig != nil && data.CacheMemoryConfig.Enabled {
+		yaml.WriteString("          \n")
+		yaml.WriteString("          ---\n")
+		yaml.WriteString("          \n")
+		yaml.WriteString("          ## Cache Folder Available\n")
+		yaml.WriteString("          \n")
+		yaml.WriteString("          You have access to a persistent cache folder at `/tmp/cache-memory/` where you can read and write files to create memories and store information.\n")
+		yaml.WriteString("          \n")
+		yaml.WriteString("          - **Read/Write Access**: You can freely read from and write to any files in this folder\n")
+		yaml.WriteString("          - **Persistence**: Files in this folder persist across workflow runs via GitHub Actions cache\n")
+		yaml.WriteString("          - **Last Write Wins**: If multiple processes write to the same file, the last write will be preserved\n")
+		yaml.WriteString("          - **File Share**: Use this as a simple file share - organize files as you see fit\n")
+		yaml.WriteString("          \n")
+		yaml.WriteString("          Examples of what you can store:\n")
+		yaml.WriteString("          - `/tmp/cache-memory/notes.txt` - general notes and observations\n")
+		yaml.WriteString("          - `/tmp/cache-memory/preferences.json` - user preferences and settings\n")
+		yaml.WriteString("          - `/tmp/cache-memory/history.log` - activity history and logs\n")
+		yaml.WriteString("          - `/tmp/cache-memory/state/` - organized state files in subdirectories\n")
+		yaml.WriteString("          \n")
+		yaml.WriteString("          Feel free to create, read, update, and organize files in this folder as needed for your tasks.\n")
+	}
+
 	if data.SafeOutputs != nil {
 		// Add output instructions for all engines (GITHUB_AW_SAFE_OUTPUTS functionality)
 		yaml.WriteString("          \n")
@@ -3846,14 +3874,6 @@ func (c *Compiler) extractCacheMemoryConfig(tools map[string]any) *CacheMemoryCo
 			}
 		}
 
-		// Parse custom docker image (deprecated)
-		if dockerImage, exists := configMap["docker-image"]; exists {
-			if dockerImageStr, ok := dockerImage.(string); ok {
-				config.DockerImage = dockerImageStr
-				// Note: docker-image is deprecated and ignored when using npx
-			}
-		}
-
 		// Parse retention days
 		if retentionDays, exists := configMap["retention-days"]; exists {
 			if retentionDaysInt, ok := retentionDays.(int); ok {
diff --git a/pkg/workflow/custom_engine.go b/pkg/workflow/custom_engine.go
index 82bf621ef4..fd57c7bf95 100644
--- a/pkg/workflow/custom_engine.go
+++ b/pkg/workflow/custom_engine.go
@@ -244,32 +244,13 @@ func (e *CustomEngine) renderCustomMCPConfig(yaml *strings.Builder, toolName str
 
 // renderCacheMemoryMCPConfig generates the Memory MCP server configuration using shared logic
 // Uses Docker-based @modelcontextprotocol/server-memory setup
+// renderCacheMemoryMCPConfig handles cache-memory configuration without MCP server mounting
+// Cache-memory is now a simple file share, not an MCP server
 func (e *CustomEngine) renderCacheMemoryMCPConfig(yaml *strings.Builder, isLast bool, workflowData *WorkflowData) {
-	// Determine Docker image to use
-	dockerImage := "mcp/memory" // default from official documentation
-	if workflowData.CacheMemoryConfig != nil && workflowData.CacheMemoryConfig.DockerImage != "" {
-		dockerImage = workflowData.CacheMemoryConfig.DockerImage
-	}
-
-	yaml.WriteString("              \"memory\": {\n")
-	yaml.WriteString("                \"command\": \"docker\",\n")
-	yaml.WriteString("                \"args\": [\n")
-	yaml.WriteString("                  \"run\",\n")
-	yaml.WriteString("                  \"-i\",\n")
-	yaml.WriteString("                  \"--rm\",\n")
-	yaml.WriteString("                  \"-v\",\n")
-	yaml.WriteString("                  \"/tmp/cache-memory:/app/dist\",\n")
-	fmt.Fprintf(yaml, "                  \"%s\"\n", dockerImage)
-	yaml.WriteString("                ],\n")
-	yaml.WriteString("                \"env\": {\n")
-	yaml.WriteString("                  \"MEMORY_FILE_PATH\": \"/app/dist/memory.json\"\n")
-	yaml.WriteString("                }\n")
-
-	if isLast {
-		yaml.WriteString("              }\n")
-	} else {
-		yaml.WriteString("              },\n")
-	}
+	// Cache-memory no longer uses MCP server mounting
+	// The cache folder is available as a simple file share at /tmp/cache-memory/
+	// The folder is created by the cache step and is accessible to all tools
+	// No MCP configuration is needed for simple file access
 }
 
 // renderSafeOutputsMCPConfig generates the Safe Outputs MCP server configuration