MCP GoGrab AI — Complete Tools Reference

17 unified gateways consolidating 90+ individual tools into a single, powerful MCP server.

Quick Reference

1. preflight

Purpose: Get intelligent context (~100 tokens) before starting any task. Returns session summaries, past decisions, and warnings. Call this once at the start of every task.

preflight({ task: "refactor auth module" })
preflight({ task: "fix drag-drop", sessionNumber: 42, subSessionNumber: 2 })

2. log_decision

Purpose: Record a development decision for learning and reinforcement. Auto-assigns to the most relevant agent and auto-detects the current session.

log_decision({
  type: "architecture",
  question: "State management for chat sessions",
  chosen: "Angular signals with computed merging",
  reasoning: "Simpler than NGXS for local UI state, avoids race conditions with autoFlashState",
  confidence: 0.9,
  options: ["NGXS store", "BehaviorSubject", "Angular signals"],
  files: ["libs/chat/chat-state.service.ts"]
})

3. find_similar

Purpose: Search past decisions by natural language query. Use after preflight when you need deeper context.

find_similar({ query: "CDK drag drop implementation" })
find_similar({ query: "auth token handling", type: "bugfix", limit: 5 })

4. session — Unified Session Management

Purpose: Consolidates 7 session tools into one gateway. Manages session lifecycle, validation, and platform detection.

session({ action: "get_info" })
session({ action: "set", sessionNumber: 42, subSessionNumber: 2 })
session({ action: "summary" })
session({ action: "validate", workspace: "editor2" })

5. agent — Multi-Agent Collaboration

Purpose: Consolidates 12 agent tools. Manages handoffs, messaging, RPC calls, and agent discovery.

Agent types: browser, code, any

Handoff response actions: acknowledge, start, complete, reject

agent({ action: "identity" })
agent({ action: "discover", task: "UI testing", requiredCapabilities: ["canBrowse"] })
agent({ action: "handoff_create", toAgentType: "browser", summary: "Test login flow", tasks: [{ description: "Verify login" }] })
agent({ action: "rpc", targetAgent: "browser-001", method: "navigate", params: { url: "https://app.example.com" } })

6. outcome — Decision Outcome Tracking

Purpose: Consolidates 3 outcome tools. Reports results for reinforcement learning loops.

Module outcomes: success, partial, failed

outcome({ action: "report", decisionId: "abc123", wasSuccessful: true, feedback: "Worked perfectly" })
outcome({ action: "track", decisionId: "abc123", triggerType: "git_commit" })
outcome({ action: "module", moduleName: "user-management", outcome: "success" })

7. stats — System Statistics

Purpose: Consolidates stats tools. Tracks tool usage, git activity, and anti-patterns.

stats({ action: "basic" })
stats({ action: "enhanced" })
stats({ action: "tools" })
stats({ action: "capture", type: "user_action", data: { event: "login" } })

8. convention — Codebase Conventions

Purpose: Manage learned codebase conventions. Read existing rules or add new ones.

convention({ action: "get" })
convention({ action: "get", category: "angular" })
convention({ action: "add", category: "testing", rule: "Use vitest over jest", examples: ["describe('...')"], source: "CLAUDE.md" })

9. intelligence — AI Super-Intelligence

Purpose: The most powerful gateway — consolidates 29 AI tools across 8 categories. Includes intelligent auto-chaining (e.g., simulate auto-recalls relevant episodes first).

Categories & Actions

Super Intelligence — Episodic memory and reasoning

Self-Healing — System health & auto-repair

Calibration — Confidence tuning

Evolution — Agent growth tracking

Code DNA — Codebase fingerprinting

Time Machine — Decision archaeology

Collective Intelligence — Multi-user learning

Utility

intelligence({ action: "recall", situation: "module integration issue" })
intelligence({ action: "simulate", simulateAction: "refactor auth", context: "need JWT", chain: true })
intelligence({ action: "diagnose" })
intelligence({ action: "calibrate", rawConfidence: 0.8, domain: "angular" })
intelligence({ action: "trace", sessionNumber: 42 })
intelligence({ action: "predict_best", taskType: "refactor", taskDescription: "Migrate state to signals" })

10. code — Code Operations

Purpose: AI-powered code review, test generation, validation, and requirements analysis.

code({ action: "review", target: "src/app/component.ts", reviewType: "security" })
code({ action: "test_gen", target: "src/app/service.ts", testType: "unit", framework: "vitest" })
code({ action: "validate", modulePath: "libs/core/utilities", strictMode: true })
code({ action: "analyze", description: "User management module", features: ["CRUD", "roles", "permissions"] })

11. platform — Platform Operations

Purpose: Query and mutate the Space Builder platform state.

Operations: create, update, delete

platform({ action: "describe" })
platform({ action: "query", query: "active modules" })
platform({ action: "mutate", operation: "create", entity: "module", data: { name: "user-mgmt" } })
platform({ action: "automate", workflow: "deploy_module" })

12. browser — Browser Automation

Purpose: Consolidates 9 browser tools. Full Playwright-powered browser control.

browser({ action: "navigate", url: "https://example.com" })
browser({ action: "click", selector: "#login-button" })
browser({ action: "fill", selector: "#email", value: "[email protected]" })
browser({ action: "screenshot", fullPage: true })
browser({ action: "extract", selector: ".article-content" })

13. desktop — Desktop Connectivity

Purpose: Consolidates 7 desktop tools. Read/write files, execute commands, clipboard access.

desktop({ action: "read_file", path: "/Users/me/config.json" })
desktop({ action: "write_file", path: "/tmp/output.txt", content: "Hello" })
desktop({ action: "list_files", path: "/Users/me/projects", recursive: true })
desktop({ action: "execute", command: "npm run build" })
desktop({ action: "clipboard_get" })

14. task — Task Runner

Purpose: Consolidates 6 task/workflow tools. Schedule, execute, and orchestrate multi-step workflows.

task({ action: "task_create", name: "build", command: "npm run build" })
task({ action: "task_execute", taskId: "build-123" })
task({ action: "task_list" })
task({ action: "workflow_create", name: "deploy", steps: [{ command: "build" }, { command: "deploy" }] })
task({ action: "workflow_run", workflowId: "deploy-456" })

15. research — Web & Docs Search

Purpose: Search the web (Tavily API) and local documentation.

research({ action: "web_search", query: "Angular 20 signals best practices" })
research({ action: "docs_search", query: "module configuration" })

16. computer_use — Autonomous Computer Control

Purpose: Goal-oriented computer control via an observe → think → act vision loop. Claude Vision analyzes screenshots to autonomously complete tasks.

Target types: browser, desktop, hybrid

computer_use({ action: "execute", description: "Navigate to google.com and search for GoGrab AI", targetType: "browser" })
computer_use({ action: "execute", description: "Fill out login form", startUrl: "https://app.example.com/login", targetType: "browser" })
computer_use({ action: "sessions" })
computer_use({ action: "audit", sessionId: "abc123" })

17. skill — Skill Registry

Purpose: Execute modular skills across browser, desktop, computer-use, and git domains.

Built-in Skills

skill({ action: "list" })
skill({ action: "execute", skillId: "git-operations", skillAction: "status", params: { cwd: "/project" } })
skill({ action: "execute", skillId: "browser-automation", skillAction: "navigate", params: { url: "https://example.com" } })
skill({ action: "search", searchAction: "screenshot" })

Recommended Workflow

1. preflight({ task: "..." })          ← Context + warnings
2. find_similar({ query: "..." })      ← Deeper history if needed
3. convention({ action: "get" })       ← Check coding standards
4. ... do the work ...
5. log_decision({ ... })               ← Record what you decided
6. outcome({ action: "report", ... })  ← Report the result

Generated from mcp-gograb-ai MCP server — 17 gateways, 90+ consolidated tools.