@mohammadhprp/system-prompt 0.10.1

package/framework/mcps/playwright-mcp/capabilities.md

@@ -0,0 +1,149 @@
+# Playwright MCP Capabilities
+
+## What It Can Do
+
+Tools are organized by capability category. Core tools are always available; others require opt-in via `--caps`.
+
+### Core Automation
+
+| Tool | Description | Read-only |
+| --- | --- | --- |
+| `browser_navigate` | Navigate to a URL | No |
+| `browser_navigate_back` | Go back to previous page in history | No |
+| `browser_click` | Perform click on a web page | No |
+| `browser_hover` | Hover over element on page | No |
+| `browser_type` | Type text into editable element | No |
+| `browser_fill_form` | Fill multiple form fields | No |
+| `browser_select_option` | Select an option in a dropdown | No |
+| `browser_drag` | Perform drag and drop between two elements | No |
+| `browser_drop` | Drop files or data onto an element | No |
+| `browser_press_key` | Press a key on the keyboard | No |
+| `browser_snapshot` | Capture accessibility snapshot of the current page | Yes |
+| `browser_take_screenshot` | Take a screenshot of the page or element | Yes |
+| `browser_evaluate` | Evaluate JavaScript expression on page or element | No |
+| `browser_console_messages` | Returns all console messages | Yes |
+| `browser_network_requests` | List network requests since page load | Yes |
+| `browser_network_request` | Get full details of a specific network request | Yes |
+| `browser_file_upload` | Upload one or multiple files | No |
+| `browser_handle_dialog` | Accept or dismiss a browser dialog | No |
+| `browser_wait_for` | Wait for text to appear/disappear or a set time | No |
+| `browser_close` | Close the current page | No |
+| `browser_resize` | Resize the browser window | No |
+| `browser_run_code_unsafe` | Run arbitrary Playwright code (RCE-equivalent) | No |
+
+### Tab Management
+
+| Tool | Description | Read-only |
+| --- | --- | --- |
+| `browser_tabs` | List, create, close, or select a browser tab | No |
+
+### Network (opt-in via `--caps=network`)
+
+| Tool | Description | Read-only |
+| --- | --- | --- |
+| `browser_route` | Set up route to mock network requests | No |
+| `browser_route_list` | List all active network routes | Yes |
+| `browser_unroute` | Remove network routes | No |
+| `browser_network_state_set` | Set browser online/offline | No |
+
+### Storage (opt-in via `--caps=storage`)
+
+| Tool | Description | Read-only |
+| --- | --- | --- |
+| `browser_cookie_list` | List all cookies | Yes |
+| `browser_cookie_get` | Get a specific cookie by name | Yes |
+| `browser_cookie_set` | Set a cookie with optional flags | No |
+| `browser_cookie_delete` | Delete a specific cookie | No |
+| `browser_cookie_clear` | Clear all cookies | No |
+| `browser_localstorage_list` | List all localStorage key-value pairs | Yes |
+| `browser_localstorage_get` | Get a localStorage item by key | Yes |
+| `browser_localstorage_set` | Set a localStorage item | No |
+| `browser_localstorage_delete` | Delete a localStorage item | No |
+| `browser_localstorage_clear` | Clear all localStorage | No |
+| `browser_sessionstorage_list` | List all sessionStorage key-value pairs | Yes |
+| `browser_sessionstorage_get` | Get a sessionStorage item by key | Yes |
+| `browser_sessionstorage_set` | Set a sessionStorage item | No |
+| `browser_sessionstorage_delete` | Delete a sessionStorage item | No |
+| `browser_sessionstorage_clear` | Clear all sessionStorage | No |
+| `browser_storage_state` | Save storage state (cookies, localStorage) to file | Yes |
+| `browser_set_storage_state` | Restore storage state from a file | No |
+
+### DevTools (opt-in via `--caps=devtools`)
+
+| Tool | Description | Read-only |
+| --- | --- | --- |
+| `browser_annotate` | Open Playwright Dashboard in annotation mode | Yes |
+| `browser_hide_highlight` | Remove highlight overlay from an element | No |
+| `browser_highlight` | Highlight an element on the page | No |
+| `browser_inspect` | Inspect element's computed styles and attributes | Yes |
+| `browser_measure` | Perform pixel measurement between elements | Yes |
+
+### Vision (opt-in via `--caps=vision`)
+
+| Tool | Description | Read-only |
+| --- | --- | --- |
+| `browser_click_position` | Click at specific x/y coordinates | No |
+| `browser_drag_position` | Drag from one position to another | No |
+| `browser_hover_position` | Hover at specific x/y coordinates | No |
+
+### PDF (opt-in via `--caps=pdf`)
+
+| Tool | Description | Read-only |
+| --- | --- | --- |
+| `browser_pdf` | Print current page to PDF | Yes |
+
+### Configuration (opt-in via `--caps=config`)
+
+| Tool | Description | Read-only |
+| --- | --- | --- |
+| `browser_get_config` | Get the final resolved config after merging CLI, env, and config file | Yes |
+
+## What It Cannot Do
+
+- It cannot work without Node.js 18+ and a supported browser engine.
+- It does **not** provide a security boundary — the `browser_run_code_unsafe` tool is RCE-equivalent. Rely on client-level permissions for security.
+- It cannot access local files outside the workspace unless `--allow-unrestricted-file-access` is set.
+- Vision features require the `--caps=vision` flag.
+- The Docker image only supports headless Chromium at this time.
+- It cannot persist data across isolated sessions when `--isolated` is used.
+
+## Best Practices
+
+- Use `--headless` for CI and automated workflows.
+- Use `--isolated` for concurrent agent sessions to avoid profile conflicts.
+- Use `--caps` to enable only the tool categories you need — keeps the tool surface small.
+- Use `--viewport-size` to set a consistent browser window size for reproducible snapshots.
+- Use `--device` to emulate specific mobile devices.
+- Use `--user-data-dir` with a persistent profile for logged-in sessions across restarts.
+- Use the `--config` file for complex setups instead of long CLI arg lists.
+- Do **not** use `browser_run_code_unsafe` unless the client has restricted tool permissions; it is equivalent to arbitrary code execution.
+
+## Common Workflows
+
+### Navigate and inspect a page
+
+1. Call `browser_navigate` to open a URL.
+2. Call `browser_snapshot` to capture the accessibility tree.
+3. Call `browser_console_messages` to check for errors.
+4. Call `browser_network_requests` to inspect network activity.
+
+### Fill and submit a form
+
+1. Call `browser_navigate` to open the page.
+2. Call `browser_snapshot` to see the form structure.
+3. Call `browser_fill_form` or `browser_type` to enter values.
+4. Call `browser_click` to press the submit button.
+
+### Debug a failing test
+
+1. Call `browser_navigate` to the test URL.
+2. Call `browser_snapshot` to verify page state.
+3. Call `browser_evaluate` to run custom inspection scripts.
+4. Call `browser_take_screenshot` to capture visual state.
+
+### Mock API responses
+
+1. Enable `--caps=network` in server args.
+2. Call `browser_route` to intercept a URL pattern and return mock data.
+3. Call `browser_navigate` to the page that uses the API.
+4. Call `browser_network_requests` to verify the mock was used.

package/framework/mcps/playwright-mcp/configs/opencode.json

@@ -0,0 +1,9 @@
+{
+  "mcp": {
+    "playwright": {
+      "type": "local",
+      "enabled": true,
+      "command": ["npx", "@playwright/mcp@latest"]
+    }
+  }
+}

package/framework/mcps/playwright-mcp/install.md

@@ -0,0 +1,136 @@
+# Playwright MCP Installation
+
+## Requirements
+
+- Node.js 18 or newer available on `PATH`.
+- An MCP-compatible AI client.
+- Browser binaries are downloaded automatically on first run (Chromium by default).
+
+## Installation
+
+The server is used directly with `npx` — no build step required.
+
+```bash
+npx @playwright/mcp@latest
+```
+
+## Configuration
+
+| Field | Value |
+| --- | --- |
+| Server name | `playwright` |
+| Command | `npx` |
+| Args | `@playwright/mcp@latest` |
+
+### Config file examples
+
+Use the examples in [`configs/`](./configs/) as client-specific starting points:
+
+| Client | Example | Typical location |
+| --- | --- | --- |
+| OpenCode | [`configs/opencode.json`](./configs/opencode.json) | OpenCode MCP configuration. |
+
+### OpenCode
+
+Copy [`configs/opencode.json`](./configs/opencode.json) into your OpenCode project root or user config directory and restart the OpenCode agent.
+
+### Common flags
+
+Pass these via the `args` array:
+
+| Flag | Description |
+| --- | --- |
+| `--headless` | Run browser in headless (no UI) mode |
+| `--browser=firefox` | Use Firefox instead of Chromium (also: `webkit`, `chrome`, `msedge`) |
+| `--caps=vision` | Enable vision capability for coordinate-based interactions |
+| `--caps=pdf` | Enable PDF generation capability |
+| `--caps=devtools` | Enable DevTools capability |
+| `--caps=network` | Enable network mocking/routing capability |
+| `--caps=storage` | Enable cookie/localStorage management capability |
+| `--isolated` | Keep browser profile in memory, do not save to disk |
+| `--port=8931` | Listen on a port for SSE transport (standalone server) |
+| `--user-data-dir=/path` | Custom user data directory for persistent profile |
+| `--device="iPhone 15"` | Emulate a mobile device |
+| `--no-sandbox` | Disable sandbox (needed in some Docker/CI environments) |
+| `--viewport-size=1280x720` | Set browser viewport size |
+
+Headless Chromium config example:
+
+```json
+{
+  "mcp": {
+    "playwright": {
+      "type": "local",
+      "enabled": true,
+      "command": ["npx", "@playwright/mcp@latest", "--headless"]
+    }
+  }
+}
+```
+
+### Docker
+
+Playwright MCP supports Docker for headless Chromium:
+
+```json
+{
+  "mcp": {
+    "playwright": {
+      "type": "local",
+      "enabled": true,
+      "command": ["docker", "run", "-i", "--rm", "--init", "--pull=always", "mcr.microsoft.com/playwright/mcp"]
+    }
+  }
+}
+```
+
+Or as a long-lived service:
+
+```bash
+docker run -d -i --rm --init --pull=always \
+  --entrypoint node \
+  --name playwright \
+  -p 8931:8931 \
+  mcr.microsoft.com/playwright/mcp \
+  /app/cli.js --headless --browser chromium --no-sandbox --port 8931 --host 0.0.0.0
+```
+
+## Verification
+
+Run the MCP server command directly:
+
+```bash
+npx @playwright/mcp@latest
+```
+
+Then verify from your AI client:
+
+- The `playwright` MCP server is listed.
+- The server starts without errors.
+- Tools are visible to the agent.
+
+Recommended first prompt:
+
+```
+Navigate to https://playwright.dev and take a snapshot of the page
+```
+
+## Updating
+
+The `@latest` tag ensures your client always uses the newest version. To force an update:
+
+```bash
+npx @playwright/mcp@latest
+```
+
+## Uninstalling
+
+Remove the `playwright` MCP entry from AI client configuration files.
+
+## Common Issues
+
+- **Browser does not start**: ensure browser binaries are available or use `--browser` to specify an installed browser.
+- **Sandbox errors on Linux**: add `--no-sandbox` to the args when running in Docker or CI.
+- **Server not found**: confirm the config file is in the correct location and the JSON is valid.
+- **Concurrent session conflicts**: persistent profiles cannot be shared. Use `--isolated` or a distinct `--user-data-dir` for parallel clients.
+- **Port already in use**: specify a different port with `--port`.

package/framework/mcps/playwright-mcp/troubleshooting.md

@@ -0,0 +1,154 @@
+# Playwright MCP Troubleshooting
+
+## Server won't start
+
+**Problem**
+
+The AI client reports that `playwright` failed to start.
+
+**Cause**
+
+The client cannot run `npx`, Node.js is not installed, or browser binaries are missing.
+
+**Solution**
+
+- Run `npx @playwright/mcp@latest` manually to verify the server starts.
+- Check Node.js is installed: `node --version` (requires 18+).
+- On first run, Playwright downloads browser binaries automatically. Ensure a working internet connection.
+- Check network access if the npm package has not been cached locally.
+
+## Browser does not launch
+
+**Problem**
+
+The server starts but no browser window appears.
+
+**Cause**
+
+Headless mode is enabled, or the browser binary is not found.
+
+**Solution**
+
+- If using `--headless`, the browser runs without a UI — this is expected.
+- Use `--browser` to specify an installed browser engine (chromium, firefox, webkit).
+- Set `PLAYWRIGHT_MCP_EXECUTABLE_PATH` or pass `--executable-path` to point to a custom browser binary.
+
+## Sandbox errors on Linux
+
+**Problem**
+
+The browser fails to start with sandbox-related errors on Linux.
+
+**Cause**
+
+Chrome/Chromium sandbox requires specific kernel settings not available in Docker or containerized environments.
+
+**Solution**
+
+- Pass `--no-sandbox` in the server args.
+- Or set the `PLAYWRIGHT_MCP_NO_SANDBOX` environment variable.
+- Note: disabling sandbox reduces security. Use only in trusted environments.
+
+## "Cannot find module" errors
+
+**Problem**
+
+The server fails with module resolution errors.
+
+**Cause**
+
+The npm package is corrupted or incompletely cached.
+
+**Solution**
+
+- Clear npm cache: `npm cache clean --force`.
+- Retry with `npx @playwright/mcp@latest`.
+
+## Configuration errors
+
+**Problem**
+
+The MCP server is not listed, or the client ignores the config file.
+
+**Cause**
+
+The config file is in the wrong location, the JSON is invalid, or the client expects a different key name.
+
+**Solution**
+
+- Validate the JSON with `python -m json.tool <file>` or another JSON parser.
+- Confirm the client-specific config location in the client's documentation.
+- Keep the server name as `playwright`.
+
+## Concurrent session conflicts
+
+**Problem**
+
+A second MCP client fails to start or shows stale data from another session.
+
+**Cause**
+
+The persistent profile (at `~/Library/Caches/ms-playwright/mcp-...` on macOS) can only be used by one browser instance at a time.
+
+**Solution**
+
+- Start additional clients with `--isolated` to keep profiles in memory only.
+- Or set a distinct `--user-data-dir` for each concurrent client.
+- Or change the workspace root so a different `workspace-hash` is derived.
+
+## Port already in use
+
+**Problem**
+
+The standalone MCP server (with `--port`) fails to bind.
+
+**Solution**
+
+- Specify a different port: `--port 8932`.
+- Or kill the process using the port: `lsof -ti:8931 | xargs kill`.
+
+## Docker issues
+
+**Problem**
+
+The Docker container fails to run or the browser does not work.
+
+**Cause**
+
+Docker image only supports headless Chromium. Missing kernel capabilities for browser execution.
+
+**Solution**
+
+- Ensure `--headless` and `--no-sandbox` are passed as args.
+- Use `--pull=always` to get the latest image.
+- Verify the container has the `--init` flag for proper signal handling.
+- For full control, build the Docker image locally with `docker build -t mcr.microsoft.com/playwright/mcp .`.
+
+## Performance issues
+
+**Problem**
+
+The MCP server is slow or times out frequently.
+
+**Solution**
+
+- Adjust action timeout: `--timeout-action 10000` (default 5000ms).
+- Adjust navigation timeout: `--timeout-navigation 120000` (default 60000ms).
+- Reduce snapshot depth with `--snapshot-mode=none` if you don't need the full accessibility tree.
+- Limit console output level: `--console-level=error`.
+
+## Extension mode not working
+
+**Problem**
+
+`--extension` flag does not connect to the running browser.
+
+**Cause**
+
+The Playwright browser extension is not installed, or Chrome is not running with the extension enabled.
+
+**Solution**
+
+- Install the Playwright Extension from [microsoft/playwright/packages/extension](https://github.com/microsoft/playwright/tree/main/packages/extension).
+- Ensure Chrome/Edge is running with the extension enabled before starting the MCP server.
+- This mode requires Chrome or Edge only; Firefox and WebKit are not supported.

package/framework/mcps/supabase-mcp/README.md

@@ -0,0 +1,66 @@
+# Supabase MCP
+
+## Overview
+
+Supabase MCP is a hosted, officially maintained MCP server that lets AI agents manage Supabase projects — query databases, design schema, run migrations, deploy Edge Functions, manage Auth users and Storage, and more. It uses OAuth authentication and communicates over Streamable HTTP.
+
+Official source:
+
+- [Supabase MCP Server docs](https://supabase.com/docs/guides/ai-tools/mcp)
+- [Tool reference](https://supabase.com/docs/guides/ai-tools/mcp#available-tools)
+- [Configuration options](https://supabase.com/docs/guides/ai-tools/mcp#configuration-options)
+- [GitHub: supabase/mcp](https://github.com/supabase/mcp)
+
+## Features
+
+- Execute SQL queries against your database.
+- List tables, extensions, and migrations.
+- Apply database migrations.
+- Retrieve service logs (API, Postgres, Edge Functions, Auth, Storage, Realtime).
+- Get security and performance advisors.
+- Get project API URLs and publishable keys.
+- Generate TypeScript types from database schema.
+- List, get, and deploy Edge Functions.
+- Manage Supabase projects and organizations.
+- Get cost information and confirm costs.
+- Search Supabase documentation.
+- Create, merge, list, delete, reset, and rebase database branches (paid plans, experimental).
+- List storage buckets and get/update storage configuration.
+- OAuth authentication — no manual token management.
+- Feature groups to restrict which tools are exposed.
+- Read-only mode and project-scoped mode for security.
+
+## Supported AI Clients
+
+Supabase MCP officially documents setup for Claude Code, Cursor, VS Code (GitHub Copilot), Windsurf, ChatGPT, Claude.ai, Goose, Factory, Codex, Gemini CLI, Kiro, Kimi Code, Antigravity, and OpenCode.
+
+- OpenCode
+
+## When to Use
+
+Use Supabase MCP when an AI coding agent needs to interact with Supabase projects — managing schema, querying data, deploying functions, or inspecting configuration.
+
+Good fits:
+
+- Designing tables and generating migrations from scratch.
+- Querying data for reports or debugging.
+- Deploying and listing Edge Functions.
+- Generating TypeScript types from an existing schema.
+- Managing Supabase projects (create, pause, restore).
+- Searching Supabase documentation for API guidance.
+- Creating and managing development branches.
+
+Avoid using it with production data. Supabase MCP is designed for development and testing. Use read-only mode, project scoping, and feature groups to limit blast radius.
+
+## Requirements
+
+- A Supabase account with a project.
+- An MCP-compatible AI client that supports remote HTTP (Streamable HTTP) MCP servers.
+- For branching features: a paid Supabase plan.
+
+## Related Skills
+
+Relevant skills in this repository:
+
+- [`backend-engineer`](../../skills/backend-engineer/SKILL.md): designing and managing database schemas.
+- [`backend-best-practices`](../../skills/backend-best-practices/SKILL.md): backend best practices for API design, testing, and debugging.

package/framework/mcps/supabase-mcp/capabilities.md

@@ -0,0 +1,127 @@
+# Supabase MCP Capabilities
+
+## What It Can Do
+
+25+ tools organized into 8 feature groups. All groups except Storage are enabled by default.
+
+### Database
+
+| Tool | Description |
+| --- | --- |
+| `list_tables` | List all tables in the database |
+| `list_extensions` | List available/installed Postgres extensions |
+| `list_migrations` | List database migrations |
+| `apply_migration` | Apply a database migration |
+| `execute_sql` | Execute SQL queries |
+
+### Debugging
+
+| Tool | Description |
+| --- | --- |
+| `get_logs` | Retrieve service logs (API, Postgres, Edge Functions, Auth, Storage, Realtime) |
+| `get_advisors` | Get security and performance advisors |
+
+### Development
+
+| Tool | Description |
+| --- | --- |
+| `get_project_url` | Get the API URL for a project |
+| `get_publishable_keys` | Get publishable and legacy anon API keys for a project |
+| `generate_typescript_types` | Generate TypeScript types from database schema |
+
+### Edge Functions
+
+| Tool | Description |
+| --- | --- |
+| `list_edge_functions` | List all Edge Functions |
+| `get_edge_function` | Get a specific Edge Function |
+| `deploy_edge_function` | Deploy an Edge Function |
+
+### Account Management
+
+Disabled when using `project_ref` scoping.
+
+| Tool | Description |
+| --- | --- |
+| `list_projects` | List all Supabase projects |
+| `get_project` | Get project details |
+| `create_project` | Create a new Supabase project |
+| `pause_project` | Pause a project |
+| `restore_project` | Restore a paused project |
+| `list_organizations` | List organizations |
+| `get_organization` | Get organization details |
+| `get_cost` | Get cost information for a project |
+| `confirm_cost` | Confirm cost for a project |
+
+### Docs
+
+| Tool | Description |
+| --- | --- |
+| `search_docs` | Search Supabase documentation via the Content API (hybrid semantic + keyword search) |
+
+### Branching (experimental, paid plans)
+
+| Tool | Description |
+| --- | --- |
+| `create_branch` | Create a database branch |
+| `list_branches` | List all branches |
+| `delete_branch` | Delete a branch |
+| `merge_branch` | Merge a branch into parent |
+| `reset_branch` | Reset a branch to a previous state |
+| `rebase_branch` | Rebase a branch onto parent |
+
+### Storage (disabled by default)
+
+| Tool | Description |
+| --- | --- |
+| `list_storage_buckets` | List storage buckets |
+| `get_storage_config` | Get storage configuration |
+| `update_storage_config` | Update storage configuration |
+
+## What It Cannot Do
+
+- It cannot work without OAuth authentication — no PAT required by default, but PAT is an option for CI.
+- It cannot access Supabase resources outside the authenticated user's permissions.
+- It cannot read/write files inside storage buckets (listing and config only).
+- It cannot create or manage Auth users directly (only view logs).
+- Branching requires a paid Supabase plan.
+- Storage tools are disabled by default — must be explicitly enabled via `features` parameter.
+- It cannot operate fully headless in its default OAuth mode — PAT auth required for CI.
+
+## Configuration Options
+
+URL query parameters to restrict scope and reduce blast radius:
+
+- `read_only=true` — executes all queries as a read-only Postgres user (no inserts/updates/deletes).
+- `project_ref=<id>` — scopes to a single project, disables account-level tools.
+- `features=<groups>` — comma-separated list of feature groups to enable (e.g. `database,docs`).
+
+## Security Best Practices
+
+- **Never connect to production.** Use a development project with non-production data.
+- **Use read-only mode** when connecting to any real data.
+- **Scope to one project** with `project_ref` to limit blast radius.
+- **Trim feature groups** to only what the task needs (e.g. `features=database,docs`).
+- **Keep manual tool approval on** in the MCP client — review each tool call before executing.
+
+## Common Workflows
+
+### Explore database schema
+
+1. Call `list_tables` to see available tables.
+2. Call `execute_sql` with a `SELECT` query to inspect table structure or data.
+
+### Generate TypeScript types
+
+1. Call `list_tables` to see the schema.
+2. Call `generate_typescript_types` to get typed definitions.
+
+### Deploy an Edge Function
+
+1. Call `list_edge_functions` to see existing functions.
+2. Call `deploy_edge_function` with the function name and code.
+
+### Search and apply from docs
+
+1. Call `search_docs` to find Supabase documentation on a topic.
+2. Call `execute_sql` to apply the documented approach.

package/framework/mcps/supabase-mcp/configs/opencode.json

@@ -0,0 +1,9 @@
+{
+  "mcp": {
+    "supabase": {
+      "type": "remote",
+      "url": "https://mcp.supabase.com/mcp",
+      "enabled": true
+    }
+  }
+}