@ngxtm/devkit 3.15.0 → 3.16.0

package/skills/microsoft-teams-automation/SKILL.md

@@ -0,0 +1,216 @@
+---
+name: microsoft-teams-automation
+description: "Automate Microsoft Teams tasks via Rube MCP (Composio): send messages, manage channels, create meetings, handle chats, and search messages. Always search tools first for current schemas."
+requires:
+  mcp: [rube]
+risk: unknown
+source: community
+---
+
+# Microsoft Teams Automation via Rube MCP
+
+Automate Microsoft Teams operations through Composio's Microsoft Teams toolkit via Rube MCP.
+
+## Prerequisites
+
+- Rube MCP must be connected (RUBE_SEARCH_TOOLS available)
+- Active Microsoft Teams connection via `RUBE_MANAGE_CONNECTIONS` with toolkit `microsoft_teams`
+- Always call `RUBE_SEARCH_TOOLS` first to get current tool schemas
+
+## Setup
+
+**Get Rube MCP**: Add `https://rube.app/mcp` as an MCP server in your client configuration. No API keys needed — just add the endpoint and it works.
+
+
+1. Verify Rube MCP is available by confirming `RUBE_SEARCH_TOOLS` responds
+2. Call `RUBE_MANAGE_CONNECTIONS` with toolkit `microsoft_teams`
+3. If connection is not ACTIVE, follow the returned auth link to complete Microsoft OAuth
+4. Confirm connection status shows ACTIVE before running any workflows
+
+## Core Workflows
+
+### 1. Send Channel Messages
+
+**When to use**: User wants to post a message to a Teams channel
+
+**Tool sequence**:
+1. `MICROSOFT_TEAMS_TEAMS_LIST` - List teams to find target team [Prerequisite]
+2. `MICROSOFT_TEAMS_TEAMS_LIST_CHANNELS` - List channels in the team [Prerequisite]
+3. `MICROSOFT_TEAMS_TEAMS_POST_CHANNEL_MESSAGE` - Post the message [Required]
+
+**Key parameters**:
+- `team_id`: UUID of the team (from TEAMS_LIST)
+- `channel_id`: Channel ID (from LIST_CHANNELS, format: '19:...@thread.tacv2')
+- `content`: Message text or HTML
+- `content_type`: 'text' or 'html'
+
+**Pitfalls**:
+- team_id must be a valid UUID format
+- channel_id must be in thread format (e.g., '19:abc@thread.tacv2')
+- TEAMS_LIST may paginate (~100 items/page); follow @odata.nextLink to find all teams
+- LIST_CHANNELS can return 403 if user lacks access to the team
+- Messages over ~28KB can trigger 400/413 errors; split long content
+- Throttling may return 429; use exponential backoff (1s/2s/4s)
+
+### 2. Send Chat Messages
+
+**When to use**: User wants to send a direct or group chat message
+
+**Tool sequence**:
+1. `MICROSOFT_TEAMS_CHATS_GET_ALL_CHATS` - List existing chats [Optional]
+2. `MICROSOFT_TEAMS_LIST_USERS` - Find users for new chats [Optional]
+3. `MICROSOFT_TEAMS_TEAMS_CREATE_CHAT` - Create a new chat [Optional]
+4. `MICROSOFT_TEAMS_TEAMS_POST_CHAT_MESSAGE` - Send the message [Required]
+
+**Key parameters**:
+- `chat_id`: Chat ID (from GET_ALL_CHATS or CREATE_CHAT)
+- `content`: Message content
+- `content_type`: 'text' or 'html'
+- `chatType`: 'oneOnOne' or 'group' (for CREATE_CHAT)
+- `members`: Array of member objects (for CREATE_CHAT)
+
+**Pitfalls**:
+- CREATE_CHAT requires the authenticated user as one of the members
+- oneOnOne chats return existing chat if one already exists between the two users
+- group chats require at least one member with 'owner' role
+- member user_odata_bind must use full Microsoft Graph URL format
+- Chat filter support is very limited; filter client-side when needed
+
+### 3. Create Online Meetings
+
+**When to use**: User wants to schedule a Microsoft Teams meeting
+
+**Tool sequence**:
+1. `MICROSOFT_TEAMS_LIST_USERS` - Find participant user IDs [Optional]
+2. `MICROSOFT_TEAMS_CREATE_MEETING` - Create the meeting [Required]
+
+**Key parameters**:
+- `subject`: Meeting title
+- `start_date_time`: ISO 8601 start time (e.g., '2024-08-15T10:00:00Z')
+- `end_date_time`: ISO 8601 end time (must be after start)
+- `participants`: Array of user objects with user_id and role
+
+**Pitfalls**:
+- end_date_time must be strictly after start_date_time
+- Participants require valid Microsoft user_id (GUID) values, not emails
+- This creates a standalone meeting not linked to a calendar event
+- For calendar-linked meetings, use OUTLOOK_CALENDAR_CREATE_EVENT with is_online_meeting=true
+
+### 4. Manage Teams and Channels
+
+**When to use**: User wants to list, create, or manage teams and channels
+
+**Tool sequence**:
+1. `MICROSOFT_TEAMS_TEAMS_LIST` - List all accessible teams [Required]
+2. `MICROSOFT_TEAMS_GET_TEAM` - Get details for a specific team [Optional]
+3. `MICROSOFT_TEAMS_TEAMS_LIST_CHANNELS` - List channels in a team [Optional]
+4. `MICROSOFT_TEAMS_GET_CHANNEL` - Get channel details [Optional]
+5. `MICROSOFT_TEAMS_TEAMS_CREATE_CHANNEL` - Create a new channel [Optional]
+6. `MICROSOFT_TEAMS_LIST_TEAM_MEMBERS` - List team members [Optional]
+7. `MICROSOFT_TEAMS_ADD_MEMBER_TO_TEAM` - Add a member to the team [Optional]
+
+**Key parameters**:
+- `team_id`: Team UUID
+- `channel_id`: Channel ID in thread format
+- `filter`: OData filter string (e.g., "startsWith(displayName,'Project')")
+- `select`: Comma-separated properties to return
+
+**Pitfalls**:
+- TEAMS_LIST pagination: follow @odata.nextLink in large tenants
+- Private/shared channels may be omitted unless permissions align
+- GET_CHANNEL returns 404 if team_id or channel_id is wrong
+- Always source IDs from list operations; do not guess ID formats
+
+### 5. Search Messages
+
+**When to use**: User wants to find messages across Teams chats and channels
+
+**Tool sequence**:
+1. `MICROSOFT_TEAMS_SEARCH_MESSAGES` - Search with KQL syntax [Required]
+
+**Key parameters**:
+- `query`: KQL search query (supports from:, sent:, attachments, boolean logic)
+
+**Pitfalls**:
+- Newly posted messages may take 30-60 seconds to appear in search
+- Search is eventually consistent; do not rely on it for immediate delivery confirmation
+- Use message listing tools for real-time message verification
+
+## Common Patterns
+
+### Team and Channel ID Resolution
+
+```
+1. Call MICROSOFT_TEAMS_TEAMS_LIST
+2. Find team by displayName
+3. Extract team id (UUID format)
+4. Call MICROSOFT_TEAMS_TEAMS_LIST_CHANNELS with team_id
+5. Find channel by displayName
+6. Extract channel id (19:...@thread.tacv2 format)
+```
+
+### User Resolution
+
+```
+1. Call MICROSOFT_TEAMS_LIST_USERS
+2. Filter by displayName or email
+3. Extract user id (UUID format)
+4. Use for meeting participants, chat members, or team operations
+```
+
+### Pagination
+
+- Teams/Users: Follow @odata.nextLink URL for next page
+- Chats: Auto-paginates up to limit; use top for page size (max 50)
+- Use `top` parameter to control page size
+- Continue until @odata.nextLink is absent
+
+## Known Pitfalls
+
+**Authentication and Permissions**:
+- Different operations require different Microsoft Graph permissions
+- 403 errors indicate insufficient permissions or team access
+- Some operations require admin consent in the Azure AD tenant
+
+**ID Formats**:
+- Team IDs: UUID format (e.g., '87b0560f-fc0d-4442-add8-b380ca926707')
+- Channel IDs: Thread format (e.g., '19:abc123@thread.tacv2')
+- Chat IDs: Various formats (e.g., '19:meeting_xxx@thread.v2')
+- User IDs: UUID format
+- Never guess IDs; always resolve from list operations
+
+**Rate Limits**:
+- Microsoft Graph enforces throttling
+- 429 responses include Retry-After header
+- Keep requests to a few per second
+- Batch operations help reduce total request count
+
+**Message Formatting**:
+- HTML content_type supports rich formatting
+- Adaptive cards require additional handling
+- Message size limit is approximately 28KB
+- Split long content into multiple messages
+
+## Quick Reference
+
+| Task | Tool Slug | Key Params |
+|------|-----------|------------|
+| List teams | MICROSOFT_TEAMS_TEAMS_LIST | filter, select, top |
+| Get team details | MICROSOFT_TEAMS_GET_TEAM | team_id |
+| List channels | MICROSOFT_TEAMS_TEAMS_LIST_CHANNELS | team_id, filter |
+| Get channel | MICROSOFT_TEAMS_GET_CHANNEL | team_id, channel_id |
+| Create channel | MICROSOFT_TEAMS_TEAMS_CREATE_CHANNEL | team_id, displayName |
+| Post to channel | MICROSOFT_TEAMS_TEAMS_POST_CHANNEL_MESSAGE | team_id, channel_id, content |
+| List chats | MICROSOFT_TEAMS_CHATS_GET_ALL_CHATS | user_id, limit |
+| Create chat | MICROSOFT_TEAMS_TEAMS_CREATE_CHAT | chatType, members, topic |
+| Post to chat | MICROSOFT_TEAMS_TEAMS_POST_CHAT_MESSAGE | chat_id, content |
+| Create meeting | MICROSOFT_TEAMS_CREATE_MEETING | subject, start_date_time, end_date_time |
+| List users | MICROSOFT_TEAMS_LIST_USERS | filter, select, top |
+| List team members | MICROSOFT_TEAMS_LIST_TEAM_MEMBERS | team_id |
+| Add team member | MICROSOFT_TEAMS_ADD_MEMBER_TO_TEAM | team_id, user_id |
+| Search messages | MICROSOFT_TEAMS_SEARCH_MESSAGES | query |
+| Get chat message | MICROSOFT_TEAMS_GET_CHAT_MESSAGE | chat_id, message_id |
+| List joined teams | MICROSOFT_TEAMS_LIST_USER_JOINED_TEAMS | (none) |
+
+## When to Use
+This skill is applicable to execute the workflow or actions described in the overview.

package/skills/minecraft-bukkit-pro/SKILL.md

@@ -1,12 +1,14 @@
 ---
 name: minecraft-bukkit-pro
-description: Master Minecraft server plugin development with Bukkit, Spigot, and
+description: "Master Minecraft server plugin development with Bukkit, Spigot, and"
   Paper APIs. Specializes in event-driven architecture, command systems, world
   manipulation, player management, and performance optimization. Use PROACTIVELY
   for plugin architecture, gameplay mechanics, server-side features, or
   cross-version compatibility.
 metadata:
   model: opus
+risk: unknown
+source: community
 ---
 
 ## Use this skill when

package/skills/miro-automation/SKILL.md

@@ -0,0 +1,210 @@
+---
+name: miro-automation
+description: "Automate Miro tasks via Rube MCP (Composio): boards, items, sticky notes, frames, sharing, connectors. Always search tools first for current schemas."
+requires:
+  mcp: [rube]
+risk: unknown
+source: community
+---
+
+# Miro Automation via Rube MCP
+
+Automate Miro whiteboard operations through Composio's Miro toolkit via Rube MCP.
+
+## Prerequisites
+
+- Rube MCP must be connected (RUBE_SEARCH_TOOLS available)
+- Active Miro connection via `RUBE_MANAGE_CONNECTIONS` with toolkit `miro`
+- Always call `RUBE_SEARCH_TOOLS` first to get current tool schemas
+
+## Setup
+
+**Get Rube MCP**: Add `https://rube.app/mcp` as an MCP server in your client configuration. No API keys needed — just add the endpoint and it works.
+
+
+1. Verify Rube MCP is available by confirming `RUBE_SEARCH_TOOLS` responds
+2. Call `RUBE_MANAGE_CONNECTIONS` with toolkit `miro`
+3. If connection is not ACTIVE, follow the returned auth link to complete Miro OAuth
+4. Confirm connection status shows ACTIVE before running any workflows
+
+## Core Workflows
+
+### 1. List and Browse Boards
+
+**When to use**: User wants to find boards or get board details
+
+**Tool sequence**:
+1. `MIRO_GET_BOARDS2` - List all accessible boards [Required]
+2. `MIRO_GET_BOARD` - Get detailed info for a specific board [Optional]
+
+**Key parameters**:
+- `query`: Search term to filter boards by name
+- `sort`: Sort by 'default', 'last_modified', 'last_opened', 'last_created', 'alphabetically'
+- `limit`: Number of results per page (max 50)
+- `offset`: Pagination offset
+- `board_id`: Specific board ID for detailed retrieval
+
+**Pitfalls**:
+- Pagination uses offset-based approach, not cursor-based
+- Maximum 50 boards per page; iterate with offset for full list
+- Board IDs are long alphanumeric strings; always resolve by search first
+
+### 2. Create Boards and Items
+
+**When to use**: User wants to create a new board or add items to an existing board
+
+**Tool sequence**:
+1. `MIRO_CREATE_BOARD` - Create a new empty board [Optional]
+2. `MIRO_CREATE_STICKY_NOTE_ITEM` - Add sticky notes to a board [Optional]
+3. `MIRO_CREATE_FRAME_ITEM2` - Add frames to organize content [Optional]
+4. `MIRO_CREATE_ITEMS_IN_BULK` - Add multiple items at once [Optional]
+
+**Key parameters**:
+- `name` / `description`: Board name and description (for CREATE_BOARD)
+- `board_id`: Target board ID (required for all item creation)
+- `data`: Content object with `content` field for sticky note text
+- `style`: Styling object with `fillColor` for sticky note color
+- `position`: Object with `x` and `y` coordinates
+- `geometry`: Object with `width` and `height`
+
+**Pitfalls**:
+- `board_id` is required for ALL item operations; resolve via GET_BOARDS2 first
+- Sticky note colors use hex codes (e.g., '#FF0000') in the `fillColor` field
+- Position coordinates use the board's coordinate system (origin at center)
+- BULK create has a maximum items-per-request limit; check current schema
+- Frame items require `geometry` with both width and height
+
+### 3. Browse and Manage Board Items
+
+**When to use**: User wants to view, find, or organize items on a board
+
+**Tool sequence**:
+1. `MIRO_GET_BOARD_ITEMS` - List all items on a board [Required]
+2. `MIRO_GET_CONNECTORS2` - List connections between items [Optional]
+
+**Key parameters**:
+- `board_id`: Target board ID (required)
+- `type`: Filter by item type ('sticky_note', 'shape', 'text', 'frame', 'image', 'card')
+- `limit`: Number of items per page
+- `cursor`: Pagination cursor from previous response
+
+**Pitfalls**:
+- Results are paginated; follow `cursor` until absent for complete item list
+- Item types must match Miro's predefined types exactly
+- Large boards may have thousands of items; use type filtering to narrow results
+- Connectors are separate from items; use GET_CONNECTORS2 for relationship data
+
+### 4. Share and Collaborate on Boards
+
+**When to use**: User wants to share a board with team members or manage access
+
+**Tool sequence**:
+1. `MIRO_GET_BOARDS2` - Find the board to share [Prerequisite]
+2. `MIRO_SHARE_BOARD` - Share the board with users [Required]
+3. `MIRO_GET_BOARD_MEMBERS` - Verify current board members [Optional]
+
+**Key parameters**:
+- `board_id`: Board to share (required)
+- `emails`: Array of email addresses to invite
+- `role`: Access level ('viewer', 'commenter', 'editor')
+- `message`: Optional invitation message
+
+**Pitfalls**:
+- Email addresses must be valid; invalid emails cause the entire request to fail
+- Role must be one of the predefined values; case-sensitive
+- Sharing with users outside the organization may require admin approval
+- GET_BOARD_MEMBERS returns all members including the owner
+
+### 5. Create Visual Connections
+
+**When to use**: User wants to connect items on a board with lines or arrows
+
+**Tool sequence**:
+1. `MIRO_GET_BOARD_ITEMS` - Find items to connect [Prerequisite]
+2. `MIRO_GET_CONNECTORS2` - View existing connections [Optional]
+
+**Key parameters**:
+- `board_id`: Target board ID
+- `startItem`: Object with `id` of the source item
+- `endItem`: Object with `id` of the target item
+- `style`: Connector style (line type, color, arrows)
+
+**Pitfalls**:
+- Both start and end items must exist on the same board
+- Item IDs are required for connections; resolve via GET_BOARD_ITEMS first
+- Connector styles vary; check available options in schema
+- Self-referencing connections (same start and end) are not allowed
+
+## Common Patterns
+
+### ID Resolution
+
+**Board name -> Board ID**:
+```
+1. Call MIRO_GET_BOARDS2 with query=board_name
+2. Find board by name in results
+3. Extract id field
+```
+
+**Item lookup on board**:
+```
+1. Call MIRO_GET_BOARD_ITEMS with board_id and optional type filter
+2. Find item by content or position
+3. Extract item id for further operations
+```
+
+### Pagination
+
+- Boards: Use `offset` and `limit` (offset-based)
+- Board items: Use `cursor` and `limit` (cursor-based)
+- Continue until no more results or cursor is absent
+- Default page sizes vary by endpoint
+
+### Coordinate System
+
+- Board origin (0,0) is at the center
+- Positive X is right, positive Y is down
+- Items positioned by their center point
+- Use `position: {x: 0, y: 0}` for center of board
+- Frames define bounded areas; items inside inherit frame position
+
+## Known Pitfalls
+
+**Board IDs**:
+- Board IDs are required for virtually all operations
+- Always resolve board names to IDs via GET_BOARDS2 first
+- Do not hardcode board IDs; they vary by account
+
+**Item Creation**:
+- Each item type has different required fields
+- Sticky notes need `data.content` for text
+- Frames need `geometry.width` and `geometry.height`
+- Position defaults to (0,0) if not specified; items may overlap
+
+**Rate Limits**:
+- Miro API has rate limits per token
+- Bulk operations preferred over individual item creation
+- Use MIRO_CREATE_ITEMS_IN_BULK for multiple items
+
+**Response Parsing**:
+- Response data may be nested under `data` key
+- Item types determine which fields are present in response
+- Parse defensively; optional fields may be absent
+
+## Quick Reference
+
+| Task | Tool Slug | Key Params |
+|------|-----------|------------|
+| List boards | MIRO_GET_BOARDS2 | query, sort, limit, offset |
+| Get board details | MIRO_GET_BOARD | board_id |
+| Create board | MIRO_CREATE_BOARD | name, description |
+| Add sticky note | MIRO_CREATE_STICKY_NOTE_ITEM | board_id, data, style, position |
+| Add frame | MIRO_CREATE_FRAME_ITEM2 | board_id, data, geometry, position |
+| Bulk add items | MIRO_CREATE_ITEMS_IN_BULK | board_id, items |
+| Get board items | MIRO_GET_BOARD_ITEMS | board_id, type, cursor |
+| Share board | MIRO_SHARE_BOARD | board_id, emails, role |
+| Get members | MIRO_GET_BOARD_MEMBERS | board_id |
+| Get connectors | MIRO_GET_CONNECTORS2 | board_id |
+
+## When to Use
+This skill is applicable to execute the workflow or actions described in the overview.

package/skills/mixpanel-automation/SKILL.md

@@ -0,0 +1,229 @@
+---
+name: mixpanel-automation
+description: "Automate Mixpanel tasks via Rube MCP (Composio): events, segmentation, funnels, cohorts, user profiles, JQL queries. Always search tools first for current schemas."
+requires:
+  mcp: [rube]
+risk: unknown
+source: community
+---
+
+# Mixpanel Automation via Rube MCP
+
+Automate Mixpanel product analytics through Composio's Mixpanel toolkit via Rube MCP.
+
+## Prerequisites
+
+- Rube MCP must be connected (RUBE_SEARCH_TOOLS available)
+- Active Mixpanel connection via `RUBE_MANAGE_CONNECTIONS` with toolkit `mixpanel`
+- Always call `RUBE_SEARCH_TOOLS` first to get current tool schemas
+
+## Setup
+
+**Get Rube MCP**: Add `https://rube.app/mcp` as an MCP server in your client configuration. No API keys needed — just add the endpoint and it works.
+
+
+1. Verify Rube MCP is available by confirming `RUBE_SEARCH_TOOLS` responds
+2. Call `RUBE_MANAGE_CONNECTIONS` with toolkit `mixpanel`
+3. If connection is not ACTIVE, follow the returned auth link to complete Mixpanel authentication
+4. Confirm connection status shows ACTIVE before running any workflows
+
+## Core Workflows
+
+### 1. Aggregate Event Data
+
+**When to use**: User wants to count events, get totals, or track event trends over time
+
+**Tool sequence**:
+1. `MIXPANEL_GET_ALL_PROJECTS` - List projects to get project ID [Prerequisite]
+2. `MIXPANEL_AGGREGATE_EVENT_COUNTS` - Get event counts and aggregations [Required]
+
+**Key parameters**:
+- `event`: Event name or array of event names to aggregate
+- `from_date` / `to_date`: Date range in 'YYYY-MM-DD' format
+- `unit`: Time granularity ('minute', 'hour', 'day', 'week', 'month')
+- `type`: Aggregation type ('general', 'unique', 'average')
+- `where`: Filter expression for event properties
+
+**Pitfalls**:
+- Date format must be 'YYYY-MM-DD'; other formats cause errors
+- Event names are case-sensitive; use exact names from your Mixpanel project
+- `where` filter uses Mixpanel expression syntax (e.g., `properties["country"] == "US"`)
+- Maximum date range may be limited depending on your Mixpanel plan
+
+### 2. Run Segmentation Queries
+
+**When to use**: User wants to break down events by properties for detailed analysis
+
+**Tool sequence**:
+1. `MIXPANEL_QUERY_SEGMENTATION` - Run segmentation analysis [Required]
+
+**Key parameters**:
+- `event`: Event name to segment
+- `from_date` / `to_date`: Date range in 'YYYY-MM-DD' format
+- `on`: Property to segment by (e.g., `properties["country"]`)
+- `unit`: Time granularity
+- `type`: Count type ('general', 'unique', 'average')
+- `where`: Filter expression
+- `limit`: Maximum number of segments to return
+
+**Pitfalls**:
+- The `on` parameter uses Mixpanel property expression syntax
+- Property references must use `properties["prop_name"]` format
+- Segmentation on high-cardinality properties returns capped results; use `limit`
+- Results are grouped by the segmentation property and time unit
+
+### 3. Analyze Funnels
+
+**When to use**: User wants to track conversion funnels and identify drop-off points
+
+**Tool sequence**:
+1. `MIXPANEL_LIST_FUNNELS` - List saved funnels to find funnel ID [Prerequisite]
+2. `MIXPANEL_QUERY_FUNNEL` - Execute funnel analysis [Required]
+
+**Key parameters**:
+- `funnel_id`: ID of the saved funnel to query
+- `from_date` / `to_date`: Date range
+- `unit`: Time granularity
+- `where`: Filter expression
+- `on`: Property to segment funnel by
+- `length`: Conversion window in days
+
+**Pitfalls**:
+- `funnel_id` is required; resolve via LIST_FUNNELS first
+- Funnels must be created in Mixpanel UI first; API only queries existing funnels
+- Conversion window (`length`) defaults vary; set explicitly for accuracy
+- Large date ranges with segmentation can produce very large responses
+
+### 4. Manage User Profiles
+
+**When to use**: User wants to query or update user profiles in Mixpanel
+
+**Tool sequence**:
+1. `MIXPANEL_QUERY_PROFILES` - Search and filter user profiles [Required]
+2. `MIXPANEL_PROFILE_BATCH_UPDATE` - Update multiple user profiles [Optional]
+
+**Key parameters**:
+- `where`: Filter expression for profile properties (e.g., `properties["plan"] == "premium"`)
+- `output_properties`: Array of property names to include in results
+- `page`: Page number for pagination
+- `session_id`: Session ID for consistent pagination (from first response)
+- For batch update: array of profile updates with `$distinct_id` and property operations
+
+**Pitfalls**:
+- Profile queries return paginated results; use `session_id` from first response for consistent paging
+- `where` uses Mixpanel expression syntax for profile properties
+- BATCH_UPDATE applies operations (`$set`, `$unset`, `$add`, `$append`) to profiles
+- Batch update has a maximum number of profiles per request; chunk larger updates
+- Profile property names are case-sensitive
+
+### 5. Manage Cohorts
+
+**When to use**: User wants to list or analyze user cohorts
+
+**Tool sequence**:
+1. `MIXPANEL_COHORTS_LIST` - List all saved cohorts [Required]
+
+**Key parameters**:
+- No required parameters; returns all accessible cohorts
+- Response includes cohort `id`, `name`, `description`, `count`
+
+**Pitfalls**:
+- Cohorts are created and managed in Mixpanel UI; API provides read access
+- Cohort IDs are numeric; use exact ID from list results
+- Cohort counts may be approximate for very large cohorts
+- Cohorts can be used as filters in other queries via `where` expressions
+
+### 6. Run JQL and Insight Queries
+
+**When to use**: User wants to run custom JQL queries or insight analyses
+
+**Tool sequence**:
+1. `MIXPANEL_JQL_QUERY` - Execute a custom JQL (JavaScript Query Language) query [Optional]
+2. `MIXPANEL_QUERY_INSIGHT` - Run a saved insight query [Optional]
+
+**Key parameters**:
+- For JQL: `script` containing the JQL JavaScript code
+- For Insight: `bookmark_id` of the saved insight
+- `project_id`: Project context for the query
+
+**Pitfalls**:
+- JQL uses JavaScript-like syntax specific to Mixpanel
+- JQL queries have execution time limits; optimize for efficiency
+- Insight `bookmark_id` must reference an existing saved insight
+- JQL is a legacy feature; check Mixpanel documentation for current availability
+
+## Common Patterns
+
+### ID Resolution
+
+**Project name -> Project ID**:
+```
+1. Call MIXPANEL_GET_ALL_PROJECTS
+2. Find project by name in results
+3. Extract project id
+```
+
+**Funnel name -> Funnel ID**:
+```
+1. Call MIXPANEL_LIST_FUNNELS
+2. Find funnel by name
+3. Extract funnel_id
+```
+
+### Mixpanel Expression Syntax
+
+Used in `where` and `on` parameters:
+- Property reference: `properties["property_name"]`
+- Equality: `properties["country"] == "US"`
+- Comparison: `properties["age"] > 25`
+- Boolean: `properties["is_premium"] == true`
+- Contains: `"search_term" in properties["name"]`
+- AND/OR: `properties["country"] == "US" and properties["plan"] == "pro"`
+
+### Pagination
+
+- Event queries: Follow date-based pagination by adjusting date ranges
+- Profile queries: Use `page` number and `session_id` for consistent results
+- Funnel/cohort lists: Typically return complete results without pagination
+
+## Known Pitfalls
+
+**Date Formats**:
+- Always use 'YYYY-MM-DD' format
+- Date ranges are inclusive on both ends
+- Data freshness depends on Mixpanel ingestion delay (typically minutes)
+
+**Expression Syntax**:
+- Property references always use `properties["name"]` format
+- String values must be quoted: `properties["status"] == "active"`
+- Numeric values are unquoted: `properties["count"] > 10`
+- Boolean values: `true` / `false` (lowercase)
+
+**Rate Limits**:
+- Mixpanel API has rate limits per project
+- Large segmentation queries may time out; reduce date range or segments
+- Use batch operations where available to minimize API calls
+
+**Response Parsing**:
+- Response data may be nested under `data` key
+- Event data is typically grouped by date and segment
+- Numeric values may be returned as strings; parse explicitly
+- Empty date ranges return empty objects, not empty arrays
+
+## Quick Reference
+
+| Task | Tool Slug | Key Params |
+|------|-----------|------------|
+| List projects | MIXPANEL_GET_ALL_PROJECTS | (none) |
+| Aggregate events | MIXPANEL_AGGREGATE_EVENT_COUNTS | event, from_date, to_date, unit |
+| Segmentation | MIXPANEL_QUERY_SEGMENTATION | event, on, from_date, to_date |
+| List funnels | MIXPANEL_LIST_FUNNELS | (none) |
+| Query funnel | MIXPANEL_QUERY_FUNNEL | funnel_id, from_date, to_date |
+| Query profiles | MIXPANEL_QUERY_PROFILES | where, output_properties, page |
+| Batch update profiles | MIXPANEL_PROFILE_BATCH_UPDATE | (profile update objects) |
+| List cohorts | MIXPANEL_COHORTS_LIST | (none) |
+| JQL query | MIXPANEL_JQL_QUERY | script |
+| Query insight | MIXPANEL_QUERY_INSIGHT | bookmark_id |
+
+## When to Use
+This skill is applicable to execute the workflow or actions described in the overview.

package/skills/ml-engineer/SKILL.md

@@ -1,11 +1,13 @@
 ---
 name: ml-engineer
-description: Build production ML systems with PyTorch 2.x, TensorFlow, and
+description: "Build production ML systems with PyTorch 2.x, TensorFlow, and"
   modern ML frameworks. Implements model serving, feature engineering, A/B
   testing, and monitoring. Use PROACTIVELY for ML model deployment, inference
   optimization, or production ML infrastructure.
 metadata:
   model: inherit
+risk: unknown
+source: community
 ---
 
 ## Use this skill when

package/skills/ml-pipeline-workflow/SKILL.md

@@ -1,6 +1,8 @@
 ---
 name: ml-pipeline-workflow
-description: Build end-to-end MLOps pipelines from data preparation through model training, validation, and production deployment. Use when creating ML pipelines, implementing MLOps practices, or automating model training and deployment workflows.
+description: "Build end-to-end MLOps pipelines from data preparation through model training, validation, and production deployment. Use when creating ML pipelines, implementing MLOps practices, or automating mod..."
+risk: unknown
+source: community
 ---
 
 # ML Pipeline Workflow