QuickEngineering - Guide

QuickEngineering User Guide

QuickEngineering turns a goal into a connected flow of explicit tasks, instructions, workers, inputs, results, reviews, and publishable artifacts.

Getting Started

  1. Create a Project and choose a workspace only when the flow needs files or code.
  2. Create a Flow, then add tasks with explicit Instructions and Result names.
  3. Choose an Agent for each task and map prior results into later inputs.
  4. Add review, human, data-send, budget, or publish gates where impact requires them.
  5. Run the flow, follow Runs and Agent Queue, then inspect Artifacts and History.
Tip: Name every Result clearly so downstream mappings remain understandable.

Interface Overview

AreaPurpose
ProjectsThe .qeng source container for flows, artifacts, snapshots, imports, and trash.
Flow EditorTask graph, mappings, conditions, contracts, gates, budgets, and run entry.
Prompt WorkspacePrompts, variables, variants, matrices, evaluation, and reusable assets.
Runs and Agent QueueExecution, approvals, claimed tasks, progress, retry, review, and final state.
Artifacts and HistoryTyped outputs, versions, lineage, comments, diff, rollback, export, publish, and audit.

Task → Instructions → Agent → Result

The Flow Editor keeps each task's purpose, prompt, worker, input contract, output contract, and review boundary visible rather than burying orchestration in one chat.

  • Flow Editor: Connect tasks, inputs, results, mappings, contracts, conditions, budgets, and gates.
  • Prompt Workspace: Manage reusable prompts, variables, variants, matrices, evaluation, and versioned prompt assets.
  • Agent routing: Use System AI, local models, MCP agents, configured cloud providers, or human steps.
  • Durable Runs: Track queued, active, paused, waiting, failed, reviewed, and completed work with retry boundaries.
  • Agent Queue and MCP: Let scoped external agents claim work, receive context, and return progress and results.
  • Artifacts and review: Keep typed outputs, lineage, versions, comments, diffs, rollback, export, publish, and audit history.
Note: External agents, providers, commands, and connected services can affect files and systems. Review scopes, destinations, and outputs before approval.

Settings

Open Settings to review the following app-specific controls:

  • Agents, AI providers, connections, MCP Host, and workspace permissions
  • Approval, data-send, budget, review, and publish policies
  • Keyboard shortcuts, appearance, diagnostics, and recovery
  • Project, artifact, trigger, Knowledge, and Agent Memory behavior

Trial, Free, and Pro use the same functional feature set. Pro removes mobile ads and the daily desktop support reminder.

Keyboard Shortcuts

On Apple platforms, use Command where the app maps a displayed Ctrl shortcut to the platform primary modifier. The in-app shortcut page remains authoritative.

ShortcutAction
F2 / EnterRename or edit the selected prompt variable item
F4Toggle the selected prompt variable path mode
Tab / Right ArrowMove into the next prompt variable segment
Left ArrowMove to the previous prompt variable segment
EscapeCancel inline prompt editing
Settings > Keyboard ShortcutsReview the currently registered global actions

Tips & Tricks

Name every Result clearly so downstream mappings remain understandable.
Start with System AI and a short manual flow before adding external Agents or triggers.
Use VFS-mediated or Patch-only access when direct agent workspace access does not pass its probe.
Keep secrets as references and inspect Data Send Preview before external transmission.

Troubleshooting

ProblemWhat to check
An Agent is unavailableOpen Agents and Connections; check installation, login, delivery mode, listener, workspace, and probe status.
A Run is waitingInspect its task status, approval, data-send review, human gate, budget, or missing input.
A result cannot applyReview the artifact contract, source revision, workspace capability, and Review/Gate decision.
A background listener is not claiming tasksConfirm MCP Host, project routing, listener process, exact Agent, and current workspace path.

Privacy

  • Secrets remain references rather than expanded project text.
  • Data Send Preview shows the destination and selected payload before external transmission.
  • Agent and MCP access is scoped, capability-tested, and routed through review boundaries.
  • Audit History records decisions and results without intentionally exposing raw credentials.
Your responsibility: Use only authorized providers, workspaces, hosts, accounts, and tools; do not bypass approval, audit, or safety controls.