wolfpack-mcp 1.0.14 → 1.0.16

package/README.md

@@ -93,6 +93,15 @@ To pre-select a default team, add to your config:
 }
 ```
 
+## Project Focus
+
+The MCP server is designed to work within **one project at a time**:
+
+- The agent will identify which project you're working in before taking actions
+- Once a project is identified, all operations stay within that project
+- The agent will not perform bulk actions across multiple projects
+- To switch projects, explicitly tell Claude (e.g., "switch to the Platform project")
+
 ## API Key Permissions
 
 When creating your API key, select the permissions you need:
@@ -137,6 +146,9 @@ Verify [wolfpacks.work](https://wolfpacks.work) is accessible from your network.
 - Permissions are configurable per key—grant only what you need
 - Revoke keys anytime from Account Settings
 
+For additional security, you can **restrict API keys to specific projects** when creating them in Account Settings. This
+ensures the key can only access the projects you select, regardless of your team memberships.
+
 ## Available Tools Reference
 
 <details>

package/dist/index.js

@@ -245,7 +245,14 @@ class WolfpackMCPServer {
                 // Team tools
                 {
                     name: 'list_teams',
-                    description: 'List all teams you have access to. IMPORTANT: Call this first if you need to know the team_id for other operations. Returns team IDs, slugs, and names. Single-team users have their team auto-selected; multi-team users must specify team_id in other tool calls.',
+                    description: 'List all teams (projects) you have access to. ' +
+                        'IMPORTANT - PROJECT FOCUS RULES: ' +
+                        '1) Call this FIRST to identify which project you are working in. ' +
+                        '2) Once you identify a project, STAY WITHIN that project for all operations unless the user explicitly asks to switch projects. ' +
+                        '3) NEVER perform bulk actions across multiple projects - always work in one project at a time. ' +
+                        '4) If the user mentions a specific project, use that project for all subsequent operations. ' +
+                        '5) If unclear which project to use, ASK the user before proceeding. ' +
+                        'Returns team IDs, slugs, and names. Single-team users have their team auto-selected; multi-team users must specify team_id in other tool calls.',
                     inputSchema: {
                         type: 'object',
                         properties: {},
@@ -311,7 +318,8 @@ class WolfpackMCPServer {
                     description: 'Get a specific work item/task by ID or reference number. Returns full details including description (markdown notes). ' +
                         'Call this before updating to see current content. ' +
                         'WORKFLOW: When asked to work on an item, check its status and follow the required state transitions ' +
-                        '(pending→pull first, new→doing, blocked/ready/completed/closed→new→doing, then review when done).',
+                        '(pending→pull first, new→doing, blocked/ready/completed/closed→new→doing, then review when done). ' +
+                        'PLANNING: Check if the description contains a plan (markdown checklist). If not, APPEND one using update_work_progress - preserve all original description text and add your plan below a "---" separator.',
                     inputSchema: {
                         type: 'object',
                         properties: {
@@ -329,7 +337,9 @@ class WolfpackMCPServer {
                 },
                 {
                     name: 'update_work_progress',
-                    description: 'Update work item description/notes with your progress. WORKFLOW: 1) get_work_item to read current notes 2) Modify the markdown description with new findings/progress 3) Call this with the complete updated description. The full description replaces the existing one.',
+                    description: 'Update work item description/notes with your progress. WORKFLOW: 1) get_work_item to read current notes 2) Modify the markdown description with new findings/progress 3) Call this with the complete updated description. The full description replaces the existing one. ' +
+                        'CRITICAL: NEVER overwrite or remove the original description text. Preserve all existing content exactly as-is. Append your plan below a "---" separator. You may only modify content you previously added (your plan section). ' +
+                        'BEST PRACTICE: Append a plan with markdown checkboxes (- [ ] task). Check off completed tasks (- [x] task) as you progress. Include sections for: Plan (checklist), Approach (strategy), and Notes (discoveries/decisions).',
                     inputSchema: {
                         type: 'object',
                         properties: {
@@ -349,8 +359,8 @@ class WolfpackMCPServer {
                         '1) For "new" items: move to "doing" when starting work. ' +
                         '2) For "pending" (backlog) items: use pull_work_item first (claims and sets to "new"), then move to "doing". ' +
                         '3) For items in "blocked", "ready", "completed", or "closed": move to "new" first, then to "doing". ' +
-                        '4) When work is complete: move to "review" (NOT "completed" - that is only for deployed work). ' +
-                        '5) If work cannot be done: move to "blocked" and add a comment explaining why. ' +
+                        '4) When work is complete: move to "review" (NOT "completed" - that is only for deployed work) and add a completion comment summarizing what was done. ' +
+                        '5) If work cannot be done: move to "blocked" and add a detailed comment explaining the blocker, what was attempted, and what is needed to unblock. ' +
                         'Statuses: "pending" (backlog), "new" (to do), "doing" (in progress), "review" (work done, pending review), ' +
                         '"ready" (reviewed, awaiting deployment), "blocked", "completed" (deployed).',
                     inputSchema: {
@@ -700,7 +710,10 @@ class WolfpackMCPServer {
                 {
                     name: 'create_work_item_comment',
                     description: 'Add a comment to a work item. Requires mcp:comments:create permission. ' +
-                        'REQUIRED: When moving an item to "blocked" status, add a comment explaining why the work cannot be done.',
+                        'USE COMMENTS FOR: 1) Blocking reasons when moving to "blocked" (what failed, what was tried, what is needed). ' +
+                        '2) Completion summaries when moving to "review" (what was done, files changed, testing notes). ' +
+                        '3) Important observations or decisions that should be visible in the activity history. ' +
+                        'USE DESCRIPTION (update_work_progress) FOR: Plans, checklists, and progress tracking that need to be updated over time.',
                     inputSchema: {
                         type: 'object',
                         properties: {

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "wolfpack-mcp",
-  "version": "1.0.14",
+  "version": "1.0.16",
   "description": "MCP server for Wolfpack AI-enhanced software delivery tools",
   "main": "dist/index.js",
   "type": "module",