Docs AI Agents AI Agents Overview

AI Agents Overview

VORTΞXHQ ships with a fleet of specialised AI agents that can drive every module — API, SSH, SQL, FTP, Email, Studio — and a Global Orchestrator that chains them together.

Why Agents?

Most AI assistants are chat boxes. VORTΞXHQ agents are real autonomous workers: they pick tools, execute multi-step plans, verify the results, recover from errors, and report back with full evidence. Every agent runs through a hardened reasoning loop with permission gating, tool whitelists, memory, and cancellation support.

Agent Catalogue

AgentWhere it livesWhat it does best
Global OrchestratorFooter · AI Command CenterCross-module workflows: design → build → test → deploy. Routes to specialists.
API AgentAPI Client → API Agent drawerSend requests, sync git endpoints, generate docs, debug responses.
SSH AgentSSH Terminal → SSH Agent drawerRun remote commands, deploy code, tail logs, manage snippets.
SQL AgentSQL Client → SQL Agent drawerGenerate queries, analyse schema, fix migrations, import/export data.
FTP AgentFTP/SFTP → FTP Agent drawerNavigate workspaces, edit remote files, deploy assets safely.
Email AgentEmail → Email Agent drawerTriage inbox, summarise messages, audit HTML, cleanup test mail.
Studio AgentStudio → WorkspaceSenior full-stack developer that scaffolds projects and ships features end-to-end.
Project ArchitectSpecialist · invoked by GlobalRequirements, data models, API specs, risk analysis.
DeveloperSpecialist · invoked by GlobalCode generation, migrations, security review, git ops.
QA TesterSpecialist · invoked by GlobalTest creation, coverage, performance, security testing.
DevOps / DebugSpecialist · invoked by GlobalDeployments, CI/CD, log diagnosis, root-cause hunting.

The Real-Agent Contract

Every agent in VORTΞXHQ follows the same six-step contract — never a single tool call:

  1. PLAN — think through every step before touching anything.
  2. EXECUTE — call tools sequentially, each building on the last.
  3. VERIFY — read back results before continuing.
  4. FIX — diagnose failures and try alternatives.
  5. NARRATE — keep the chat panel updated between steps.
  6. SUMMARIZE — finish with Changed Files, Test Results, Deploy Result, Verification Checks.
Permissions. Agents only touch what you approve. Write tools fire a renderer-side confirmation (or a native dialog when configured) and the policy is configurable in Settings → AI.

Provider Agnostic

Agents work with all 7 supported providers (Vortex, OpenAI, Anthropic, DeepSeek, Groq, Ollama, Custom OpenAI-compatible). Switch provider at any time — the agent keeps its tools, memory, and active chat.

Last updated 3 hours ago

No matches.