Update English documentation to use consistent "Virtual MCPs" terminology:
- Replace all "Agent" and "Gateway" references with "Virtual MCP"
- Update endpoints: /mcp/gateway/:gatewayId → /mcp/virtual/:virtualId
- Update tool names: GATEWAY_* → VIRTUAL_*
- Establish Virtual MCPs as foundational concept

This aligns documentation with planned system architecture and
unblocks narrative documentation work in main-54j.

Closes beads-main-o55

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>

Created comprehensive documentation for Virtual MCPs following clear narrative:
- MCP compliance and benefits (protocol for action)
- Challenge of traditional MCP servers (infrastructure overhead)
- Solution: Virtual MCPs (on-demand, no server code)
- Two approaches: Composition (mix existing) + Inline (create directly)
- Hybrid capability combining both approaches
- Configuration and tool exposure strategies
- Code examples for all three patterns

Target: 700-900 words, achieved exactly 900 words
Fixed broken links, clarified terminology throughout

Closes beads-main-54j

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>

Removed exclusion mode documentation - only document inclusion mode
Removed tool exposure strategies section (planned for removal in future)
Simplified default virtual MCP explanation

Word count: 830 (within 700-900 target)

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>

Create comprehensive documentation for projects as virtual MCPs at the
organization level. Projects are organizational units that ARE virtual
MCPs themselves and can aggregate capabilities from other virtual MCPs
within their scope.

Key sections:
- Projects as virtual MCPs with composition capabilities
- Configuration hierarchy (project-level vs org-admin)
- Project scope and routing (/:org/projects/:project_id)
- CLAUDE.md and AGENTS.md support at both levels
- MCP client operation within project scope
- Practical examples and use cases

Includes planned architecture callout and comparison to Claude Code's
.claude/ configuration pattern.

Closes beads-main-ko2

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>

Create comprehensive documentation for agents concept in MCP Mesh.
Agents are single-purpose virtual MCPs designed for specific tasks,
distinguished from general-purpose projects.

Key content:
- Agent fundamentals as virtual MCPs
- Distinction between agents and projects
- Creating agents (composition, inline, hybrid modes)
- Agent configuration and AGENTS.md support
- Built-in agents and mesh MCP concept (self-referential)
- Agent types and patterns (research, development, testing, deployment)
- Single-purpose design philosophy
- Agent scope and usage

Includes 8 TypeScript configuration examples and comprehensive
documentation of 5 built-in agents (agent management, connection
management, permission management, MCP store, user management).

Note: Documents PLANNED decopilot architecture for development guidance.

Closes beads-main-589

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>

Create comprehensive documentation explaining scope as the fundamental
concept in decopilot. Covers three scope types (org, project, agent),
route-to-scope mapping, scope-sensitive operations, and virtual MCPs
at every level. Includes Mermaid diagram, route mapping table, and
practical code examples.

Closes beads-main-8f1

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>

Create comprehensive decopilot overview documentation that establishes
the fundamental architecture: decopilot only works with virtual MCPs
(editable environments). Includes comparison table for Claude Code users,
explains scope-sensitive operations (org/project/agent), task-based agent
spawning, and links to foundational concepts.

Key sections: opening hook, why virtual MCPs, Claude Code comparison,
key capabilities, how it works, examples, and next steps.

Closes beads-main-pas

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>

Create comprehensive architecture documentation covering core architectural
concepts including Virtual MCPs Only constraint, scope-based architecture,
task-based agent spawning, and tool scope sensitivity.

Key sections:
- Architectural constraint explaining why decopilot requires virtual MCPs
- Virtual MCP foundation (org/projects/agents)
- Scope-based architecture with route determination
- Task/thread system with lifecycle and UI integration
- External vs virtual MCP comparison
- Dynamic virtual MCP features and extensibility
- 4 Mermaid diagrams illustrating architecture
- 5 TypeScript code examples

Closes beads-main-4lk

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>

Create mcp-mesh/decopilot/context-management.mdx explaining how decopilot manages context window, CLAUDE.md/AGENTS.md support, and context isolation.

Key content:
- Context window composition and priority order
- CLAUDE.md support (project and org-admin levels)
- AGENTS.md support for agent-specific instructions
- Context cost management and tool filtering strategies
- Context isolation via agent spawning

Closes beads-main-s5c

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>

Create comprehensive reference documentation for decopilot's 11 built-in tools:
- Org-related tools (agent_spawn, agent_search, user_ask_question)
- Current virtual MCP tools (resource_*, prompt_*, tools_*)
- Target virtual MCP tools (discovery and enablement workflow)

Includes tool specifications, scope-sensitive behavior examples, and complete
workflows for discovering and enabling external MCP tools.

Closes beads-main-ifv

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>

Create mcp-mesh/decopilot/spawning-agents.mdx explaining how decopilot spawns agents as tasks using agent_spawn tool.

Key content:
- Task spawning overview and benefits (parallel execution, context isolation)
- Tasks vs threads terminology (UI uses "tasks", codebase uses "threads")
- agent_spawn tool signature and behavior
- Task states and lifecycle (Active, Needs Input, Successful, Failed)
- Task variants: top-level vs child (tool availability differences)
- Context isolation and inheritance patterns
- Agent selection and discovery using agent_search
- UI integration and task management
- Spawning workflows: single, parallel, sequential patterns
- Best practices for effective agent spawning

Includes 7 code examples demonstrating:
- agent_spawn usage with explicit task_id return
- agent_search for discovering agents
- Handling "Needs Input" state transitions
- Single, parallel, and sequential spawning patterns

Closes beads-main-iuu

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>

Restructure agents documentation for consistency and scannability:

- Reduce agents.mdx from 490 to 88 lines, matching projects.mdx style
- Move "Agents vs. Projects" comparison to concepts.mdx as foundational context
- Remove detailed composition examples, AGENTS.md config, and agent types
- Keep focus on essential information: what agents are, how they work, and when to use them
- Maintain all cross-references (7 page-level links remain valid)

Both projects.mdx and agents.mdx now follow parallel structure with clear, scannable sections.

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>

Replace technical implementation details with user-facing language:
- Remove endpoint reference (/:org/agents/:agent_id)
- Add "Agents provide" section with clear benefits
- Mention agents are available in sidebar under "Agents" section
- Remove "agent scope" jargon in favor of plain language
- Remove technical "direct connection" usage pattern

Focus on how users interact with agents rather than implementation details.

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>

Key corrections:
- Change "Purpose" field to "Description" and "Instructions" (actual agent fields)
- Draw parallel between agents and SKILL.md files for familiarity
- Clarify that decopilot uses description to learn when to spawn agents
- Note that agents are organization-wide
- Add cross-reference to decopilot documentation
- Update "Clear purpose statement" to "Clear description" in design principles
- Add "Detailed instructions" principle referencing SKILL.md pattern

This makes the agent concept more intuitive by connecting it to familiar SKILL.md patterns.

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>

Update product name references for consistency.

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>

Complete rewrite of decopilot documentation to emphasize MCP orchestration
and agentic workflows. Key changes:

- Rewrite overview.mdx with agentic loop focus (gather → act → verify)
- Create new quickstart.mdx with 4 practical workflow examples
- Enhance scopes.mdx with workflow examples and decision guides
- Enhance tools.mdx with patterns, use cases, and troubleshooting
- Enhance spawning-agents.mdx with advanced patterns and agent selection
- Enhance context-management.mdx with optimization strategies and budget examples
- Update navigation.ts to prioritize practical content (overview → quickstart → concepts)
- Update architecture.mdx to map agentic loop to implementation

Documentation now positions decopilot as an MCP orchestration layer rather
than a development tool, emphasizing distributed MCP coordination over HTTP
(or stdio for local servers). Virtual MCPs presented as enabler for liquid
MCP creation/modification, distributable across the web.

All beads tasks completed: main-6pc, main-tzu, main-qpk, main-ds5, main-cxu,
main-3kg, main-frw, main-dfb

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>

Refactored the Scopes section in decopilot overview to be more concise
and user-friendly:

- Condensed Organization/Project/Agent scope descriptions from 3
  subsections with bullet points into 3 brief paragraphs
- Simplified "How Scopes Layer" section to remove incorrect "Org +
  Project + Agent" combination that isn't supported
- Replaced detailed "Loading Context" section with brief transitional
  paragraph pointing to dedicated context-management page
- Added cross-references between Scopes and Context Management pages
- Maintained technical accuracy while focusing on user capabilities
  rather than implementation details

This change keeps the overview concise while detailed technical
information is available in the dedicated context-management page.

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>

Enhanced context-management.mdx with detailed scope-based loading
information:

- Replaced "What Gets Included in Context" with enhanced "How Context
  Loading Works" section with priority explanations
- Added new "Scope-Specific Context Loading" section with detailed
  breakdowns of Organization, Project, and Agent scopes
- Added context inheritance table showing what loads at each scope level
- Updated CLAUDE.md section to clarify org vs project scope behavior
- Updated AGENTS.md section to emphasize agent-scope-only loading
- Added cross-references between overview Scopes section and
  context-management technical details
- Updated Strategy 4 in optimization section with scope-specific links

This provides the technical depth on context loading that was removed
from the overview, creating better separation between user-facing
concepts (overview) and implementation details (context-management).

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>

Updated sidebar navigation and page navigation order for decopilot
section to follow the logical flow from overview's "Next Steps":

New order:
1. Overview
2. Quick Start (was missing from sidebar order)
3. Context Management (moved up)
4. Tasks and Agent Spawning
5. Tools
6. Scopes (moved down)
7. Architecture

Also improved sidebar comments to document the information architecture
structure (Getting Started → Core Concepts → Working with MCP →
Monitoring → Security → Reference).

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>

Major documentation improvements across multiple areas:

**New pages:**
- Added connections.mdx to document MCP connection management
- Added tasks-and-spawning.mdx for decopilot task lifecycle

**Removed/consolidated pages:**
- Removed connect-clients.mdx (consolidated into connections.mdx)
- Removed spawning-agents.mdx (consolidated into tasks-and-spawning.mdx)
- Removed mcp-servers.mdx (content moved to connections.mdx)

**Updated documentation:**
- AGENTS.md: Updated agent documentation structure
- Sidebar.tsx: Component updates for navigation
- introduction.mdx: Improved introduction content (en + pt-br)
- quickstart.mdx: Enhanced quick start guides (mcp-mesh + decopilot)
- overview.mdx: Updated overview content
- projects.mdx: Improved project documentation
- virtual-mcps.mdx: Enhanced virtual MCP explanations
- authorization-and-roles.mdx: Updated auth documentation
- monitoring.mdx: Improved monitoring docs
- deployment.mdx: Updated deployment guide
- creating-agents.mdx: Enhanced agent creation guide
- ui.ts: Updated i18n translations

This restructuring improves information architecture by consolidating
related content, removing redundancy, and creating clearer navigation
paths through the documentation.

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>

- Rename "Context Management" → "Context" for brevity
- Rename "Tasks and Agent Spawning" → "Tasks" for clarity
- Update all internal links and navigation references
- Maintain content structure and information hierarchy

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>

Significantly condensed and simplified the scopes documentation:
- Reduced from 444 lines to 74 lines (83% reduction)
- Removed technical implementation details (route patterns, code examples)
- Replaced "organization settings" with "organization view" and "project view"
- Simplified scope determination to two factors: view location + agent selection
- Removed Virtual MCP technical sections
- Made language more user-friendly and accessible

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>

Added comprehensive documentation for decopilot's built-in tools:
- Tool discovery & enablement (tool_search, tool_enable)
- Agent management (agent_search)
- Task management (subtask_run)
- User interaction (user_ask)
- Context access (resource_read, prompt_read)

Includes descriptions, restrictions, and expected output for each tool.

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>

Significantly condensed and simplified the tasks documentation:
- Reduced from 645 lines to 191 lines (70% reduction)
- Replaced technical "agent spawning" language with user-friendly "tasks and subtasks"
- Added clear task state explanations (In Progress, Requires Action, Success, Failed)
- Introduced task timeout behavior (30 minutes) with practical guidance
- Removed internal implementation details (threads vs tasks terminology)
- Made content more accessible to non-technical users

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>

Dramatically condensed and simplified the tools documentation:
- Reduced from 1608 lines to 103 lines (94% reduction)
- Introduced clear two-tier model: built-in tools vs scope tools
- Added concise reference table for built-in tools with task/subtask availability
- Removed extensive code examples and technical implementation details
- Simplified explanation of tool discovery and enablement workflow
- Made content significantly more scannable and user-friendly

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>

Removed the architecture page as part of documentation simplification.
Content was overly technical and redundant with other pages.

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>

- Update sidebar to reflect new documentation structure
- Reorder MCP Mesh pages: Quickstart → Overview → Core Concepts
- Add special sorting for api-reference section
- Update navigation utilities and i18n keys

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>

- Remove redundant introduction.mdx (content merged into overview)
- Remove separate authentication.mdx and authorization-and-roles.mdx pages
- Add consolidated user-management.mdx covering auth, roles, and API keys
- Remove old deploy/ directory (kubernetes-helm-chart, local-docker-compose)
- Deployment content consolidated into self-hosting section

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>

- Add overview.mdx with deployment strategy guide
- Add docker-compose.mdx with local setup instructions
- Add kubernetes.mdx with production deployment guide
- Add environment-variables.mdx reference
- Organize deployment documentation in dedicated section

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>

- Add REST API endpoints documentation
- Add MCP tools reference organized by domain
- Add authentication flows guide
- Split large api-reference.mdx into focused sections
- Improve discoverability of API documentation

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>

- Update overview.mdx with improved structure and clarity
- Update quickstart.mdx with streamlined onboarding flow
- Update concepts.mdx with clearer explanations
- Update connections.mdx, projects.mdx, agents.mdx with consistent style
- Update virtual-mcps.mdx with current architecture
- Update api-keys.mdx and api-reference.mdx to align with new structure

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>

- Standardize branding from "decocms" to "deco CMS" (with space)
- Update references in api-keys, connections, monitoring, user-management
- Update decopilot overview and virtual-mcps pages
- Improve consistency and brand presentation

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>

- Update "decocms" references to "deco CMS" in decopilot documentation
- Ensure consistent branding across all decopilot pages

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>

- Update agents.mdx, projects.mdx with "deco CMS" branding
- Update self-hosting quickstart with consistent terminology
- Complete branding consistency pass

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>

…oxy documentation

- Moved API Reference section earlier in navigation (before Self-Hosting)
- Expanded API reference with detailed Connection Proxy endpoint documentation
- Added comprehensive documentation for /mcp/:connectionId endpoint including:
  - Authentication and authorization flows
  - Request/response formats with examples
  - Streaming support and monitoring
  - Security considerations
  - Use cases and configuration options
- Removed placeholder api-reference.mdx in favor of detailed subsections

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>

- Updated resource_read examples to use correct signature (uri as first parameter, options as second)
- Updated prompt_read examples to use correct signature (name as first parameter, arguments as second)
- Updated user_ask examples with proper parameter structure
- Fixed code samples to reflect actual tool implementations

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>

…atures

- Fixed resource_read call signatures across all decopilot examples
- Changed from object parameter to (uri, options) signature
- Added lineStart and lineEnd parameters where appropriate
- Ensures code examples match actual implementation

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>

- Updated deployment documentation branding from "decocms" to "deco CMS"
- Fixed minor formatting and consistency issues in docker-compose and kubernetes docs
- Updated references in api-keys and connections documentation
- Cleaned up pt-br introduction formatting

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>

Co-authored-by: cubic-dev-ai[bot] <191113872+cubic-dev-ai[bot]@users.noreply.github.com>

- Fixed GitHub Stars component to fetch from decocms/mesh repo
- Resolved SSR hydration mismatch by:
  - Initializing state with server-matching defaults
  - Loading localStorage state after hydration in useEffect
  - Using window instead of globalThis for browser APIs
  - Adding proper client-side guards
- Refactored TreeItem component:
  - Extracted icon rendering logic into shared renderIcon function
  - Unified styling between collapsible and non-collapsible items
  - Improved semantic HTML structure (proper use of <a> vs <button>)
  - Reduced code duplication

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>

Added key props to LanguageSelector and ThemeToggle components during
client-side hydration to prevent React warnings and ensure proper
reconciliation.

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>

Update the documentation landing page to redirect users to
/en/mcp-mesh/quickstart instead of /en/introduction. This provides
a better first experience by immediately guiding users to the
quickstart guide.

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>

…use cases

Replace generic DevOps/tech examples (GitHub, deployments, Slack notifications) with concrete e-commerce scenarios (Shopify orders, inventory management, fulfillment operations, payment processing) across all documentation.

Changes:
- MCP Mesh docs: Updated agents, projects, and virtual MCPs with order fulfillment, inventory, and multi-brand store examples
- Code guides: Replaced generic tool examples (email, weather, notifications) with e-commerce tools (order confirmation, inventory check, payment processing)
- Quickstart: Added more e-commerce specific example queries

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>

Complete rewrite of the decopilot quickstart guide:
- Progressive learning approach (simple → advanced)
- Focus on practical e-commerce use cases
- Removed technical implementation details
- Added ASCII art conversation examples
- Organized by progressive feature introduction:
  1. Simple requests
  2. Subtasks for heavy work
  3. Specialist agents
  4. Context management
  5. Scope switching

Target audience: e-commerce teams (designers, content creators, PMs)
who need practical guidance without technical complexity.

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>

Update decopilot documentation examples:
- overview.mdx: Replace GitHub/Figma/Slack with Shopify/inventory/customer service
- tools.mdx: Update scope tools examples to use e-commerce integrations
- virtual-mcps.mdx: Change examples from GitHub connections to Shopify/ShipStation

Examples now focus on order fulfillment, inventory management,
and customer service workflows instead of software development tasks.

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>

…text

Update examples in built-in tools documentation:
- subtask-run.mdx: Change from GitHub/Slack to inventory/orders/shipping
- tool-enable.mdx: Replace GitHub/deployment with Shopify/inventory/shipping
- tool-search.mdx: Update from GitHub/deploy to order/inventory searches

Maintains technical accuracy while making examples more relevant
to e-commerce teams using MCP Mesh.

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>

Remove the design plan file now that the quickstart redesign
has been implemented. The plan served its purpose in guiding
the restructuring of the documentation.

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>