{"name":"Orchestrator MCP Server","description":"Bridges Claude.ai cloud sessions to PS-Tools Orchestrator","tools":[{"name":"sync_session","description":"Sync a Claude.ai session to the Orchestrator dashboard and EmpireDONE. Call this at the end of any significant work session."},{"name":"get_briefing","description":"Get the current briefing and context for a project from the Orchestrator. Use at the start of a session."},{"name":"list_projects","description":"List all active projects with their status, progress, and last activity from EmpireDONE."},{"name":"get_project","description":"Get detailed information about a specific project including tasks, progress, and recent sessions."},{"name":"add_task","description":"Add a new task/ticket to a project in EmpireDONE."},{"name":"get_memory","description":"Retrieve project memory and notes from the Orchestrator knowledge base."},{"name":"send_telegram","description":"Send a notification to Paul via Telegram."},{"name":"trigger_agent","description":"Trigger a PaulitosMAX agent to execute a real task on the VPS. Returns a ticketId to poll with get_execution_result. Use this instead of trying to execute things yourself — PaulitosMAX runs real commands with full VPS access."},{"name":"get_execution_result","description":"Poll the result of a PaulitosMAX ticket. Call this after trigger_agent to get the execution output. Poll every 10-30 seconds until status is \"done\" or \"blocked\"."},{"name":"oracle_query","description":"Send a complex question to Claude Opus 4.6 for deep analysis. Use when: architecture decisions, cross-system audits, code quality analysis, technical comparisons, anything requiring expert-level reasoning. Returns a detailed Opus analysis that you can use to build briefs and implementation plans. Complexity threshold: use when the question requires expertise beyond surface-level answers."},{"name":"verify_url","description":"Take a Playwright screenshot of a URL and send it to Paul on Telegram. Use this to verify that a deploy, design change, or fix is actually visible. Paul sees the screenshot immediately."},{"name":"get_project_recap","description":"IMPORTANT: appelle ce tool au début de chaque conv Claude Code sur un projet pour avoir instantanément le contexte. Retourne : dernière session (firstUserPrompt, lastUserPrompt, lastAssistantAction, sujets, commits, verifyStatus, openQuestion) + chantiers actifs + stats. Cross-conv + cross-device."},{"name":"get_active_chantiers","description":"Liste les chantiers actifs (pending/in_progress/blocked) d'un projet. Utile pour savoir ce qui reste à faire. Filtre needsReview par défaut."},{"name":"get_last_session","description":"Retourne le summary structuré de la dernière conv finie sur un projet (firstUserPrompt, lastUserPrompt, lastAssistantAction, commits, verifyStatus, openQuestion)."}],"connect":"Add https://mcp.ps-tools.dev to Claude.ai → Settings → MCP Servers"}