Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.qoder.com/llms.txt

Use this file to discover all available pages before exploring further.

QoderWake lets you build a team of AI employees — called Wakers — that run directly on your machine. Each Waker has a defined role, name, persona, and area of expertise. Bring them into a conversation when you need help, or set them to work autonomously on a schedule or in response to events. Their tools, memory, accessible resources, and guardrails are all configurable. This guide walks you through getting QoderWake up and running from scratch: install → open the console → create a Waker → assign tasks → manage your Wakers. Product philosophy: 7×24 digital employees — “Always awake, always working.” Product features: secure and controllable, production-ready, continuously evolving. Core concepts: Employee Role → Digital Employee (Agent) → Teams / Organization (Agent Teams). Initial roles: Backend Engineer, Frontend Engineer, Test Engineer, Product Manager, Data Analyst, Content Operator — plus Custom Roles.

​
What type of Waker do you need

The current beta covers the following roles. Pick the one that best fits your needs:
RoleJob characteristics
Backend EngineerAPI development, data modeling, service integration, performance optimization, and production stability. Follows incremental delivery, test verification, and convention-over-configuration design. Automatically analyzes bugs / issues, fixes code, and submits PRs.
Frontend EngineerFrontend interface design and implementation. Strong in component architecture, visual polish, responsive layouts, accessibility, and performance tuning. Balances UX with engineering quality through incremental delivery and evidence-based validation.
Test EngineerQuality assurance for CLI tools and web products. Focuses on test plan documents, end-to-end testing, defect reproduction, and evidence-based reports. Does not run unit tests or fix issues directly.
Product ManagerAI-native product management for software products. Goal-driven requirement lifecycle, PRD generation, user feedback analysis, competitive research, and release communication — with explicit approval gates for any external writes.
Data AnalystAI-native data analyst covering problem framing, metric definitions, data collection through DingTalk doc / sheet MCPs, data diagnosis, market context, and evidence-based recommendations.
Content OperatorAI-native content operator covering account positioning, trend insights, content calendars, Xiaohongshu post creation, visual asset briefs, post-approval publishing, comment interaction, performance review, brand compliance, and cross-platform rewrites.
CustomDefine the work style, workflow, and capabilities of any role you need. Add MCP servers, skills, and other configurations yourself.

​
Environment preparation

QoderWake supports both local and cloud deployment. Run it on a stable, always-online device with reliable network access so your Wakers can stay up 7×24.
Windows version coming soon — not yet available to the public. Install on macOS or Linux for now.

​
Local deployment

Best on a long-running local device, e.g. a Mac mini.
  • macOS: 13.0 or later.
  • Linux: Ubuntu 22.04 LTS or later, with a GUI desktop for the local Web Console.

​
Cloud deployment

Mainstream cloud desktop / cloud host environments are supported. For convenient browser login and local web management, Alibaba Cloud’s Wuying Cloud Computer (Enterprise or Personal edition) is recommended.
  • Linux: Ubuntu 22.04 or later, with a GUI desktop for the local Web Console.

​
Prerequisites

QoderWake ships with software-engineering Waker roles. Before installing, confirm:
  • git is installed.
  • Repository authentication is set up.
  • You can run clone, pull, and push against your target repos.
Both local and cloud deployments need outbound internet access for sign-in, model inference, and repository access. On enterprise networks with proxies or firewalls, allow the necessary HTTPS outbound connections in advance.

​
Step 1: Install QoderWake

​
macOS

Direct downloads: Or open Terminal and run: 
curl -fsSL https://qoder-ide.oss-ap-southeast-1.aliyuncs.com/qoderwake/install.sh | bash
The script detects your platform, downloads and installs the package, sets up the command entry on PATH, signs you in via the browser, starts the local service, and opens the Web Console. Command entry: ~/.qoderwake/bin/qoderwake.

​
Linux

Open a terminal and run: 
curl -fsSL https://qoder-ide.oss-ap-southeast-1.aliyuncs.com/qoderwake/install.sh | bash
The install script automatically:
  1. Detects your Linux CPU architecture.
  2. Downloads and verifies the matching QoderWake package for your platform.
  3. Installs the QoderWake main program and resource files.
  4. Creates the command entry and tries to add it to PATH.
  5. Opens the browser for sign-in.
  6. Starts (or restarts) the local background service after sign-in.
  7. Opens the local Web Console automatically.
Command entry: 
~/.qoderwake/bin/qoderwake
Default Web Console address: 
http://127.0.0.1:19820

​
Windows

Windows version coming soon. Stay tuned — we’ll update this page once the public release is available.

​
Step 2: Open the console

Visit http://127.0.0.1:19820/ in your browser. This is QoderWake’s local management console — Waker management, task assignment, and permission configuration all happen here. The layout:
  • Left sidebar — the “My Wakers” list with each Waker’s name and last-active time.
  • Top button — Create Waker to add a new Waker; tasks can be created from the right side.
  • Main area — click into a Waker to open its conversation view and start assigning work.
  • Right panel — created tasks and details for any currently running task.
On your first visit you’ll be prompted to sign in.

​
Step 3: Create a Waker

Click Create Waker in the left sidebar.
  1. Pick a role template. Each template comes with a persona, typical workflows, core skill set, and starter prompts. Click View details to preview before selecting.
  2. Or customize a role. If no preset fits, choose Custom Role and define the persona, expertise, and work style from scratch, then attach the skills, connectors, and permissions you want.
  3. Fill in basic info. Name, avatar, and short description. Preset roles pre-fill these fields — adjust as needed.
  4. Save. A guided setup will suggest skills to install and channels to connect; you can skip and configure later.
For a custom role, provide as much detail as possible during setup — role positioning, core responsibilities, typical workflows, required skills, and any MCP tool configurations.

​
Step 4: Put your Waker to work

Once a Waker is created, either start a conversation or set up an automated task.

​
Conversation tasks

Type your request and hit send — the Waker gets to work immediately. Mid-conversation you can:
  • Follow up or redirect — jump in any time; the Waker continues with full context.
  • Attach files or screenshots — at any message, at any stage.
  • Interrupt — stop the Waker if it’s heading the wrong way.
  • Approve or reject actions — when the Waker needs to modify files, run commands, or hit the network, it pauses for your approval.
  • Answer its questions — when it hits a decision point, pick from the prompted card.
  • Switch AI models — change models mid-conversation from the top dropdown.
  • Watch the thinking process — see reasoning in real time.
  • Review outputs — generated code patches and new files appear in the right panel as they’re produced.
  • Set the working directory — choose the local directory from the conversation footer.

​
Automated tasks

Best for “every day at 9am” or “respond whenever a new issue is filed.” Go to Waker details → Triggered Tasks → New and configure:
  • Task name and description — clear enough that the Waker can act on it directly.
  • AI model — which model to use.
  • Working directory — the local directory, repository, or project workspace where the task should run.
  • Trigger type (up to 5, mix and match):
    • Scheduled — one-off, recurring, or complex schedules (daily 9am, every Monday, first of each month).
    • Event-driven — listen for GitHub Issue / PR / comment activity.
    • Webhook — let an external system trigger the task via a fixed callback URL.
  • Max runs / expiration date (optional) — cap how many times or how long a task can run.
Once saved, the Waker runs on the cadence you set. You can pause, resume, run a one-off test, or review run history at any time.

​
Step 5: Manage your Wakers

Open any Waker’s detail page to view and adjust everything about that Waker.
SectionWhat you can do
HomeAvatar, status, days onboarded, triggered task count, conversation task count, work activity heatmap.
ProjectsBind code repositories or local directories. One project can include multiple sources, with project-level memory kept separate from the Waker’s personal memory.
Triggered TasksAdd, edit, delete, or pause automated tasks.
Conversation TasksBrowse history and resume a previous chat.
MemoryLong-term knowledge about you and the project. Sources: auto-capture during conversations, manual edits, periodic system cleanup. Stored locally and never uploaded to the cloud.
SkillsInstall from the official Qoder Skills Marketplace, upload local packages, or toggle existing skills. Built-in skills cannot be uninstalled.
ConnectorsBridges to external tools and services (GitHub, Jira, GitLab, etc.).
PermissionsTool protection — built-in security rules covering command injection, resource abuse, code execution, network abuse, sensitive file access, and privilege escalation; File protection — read/write access to local files; Built-in tools — allow / ask / disable per tool. High-risk actions trigger an approval card.

​
Waker configuration

​
Memory

Each Waker maintains its own long-term memory, accumulating knowledge about you and the project over time. Memory comes from three sources:
  • Auto-capture — information the Waker considers worth remembering is written during conversations.
  • Manual edits — open the Memory page and add or edit entries directly.
  • System cleanup — the system periodically organizes and deduplicates entries on the Waker’s behalf.
Go to Waker details → Memory to browse, search, and edit all memory entries. Memory is stored locally and never uploaded to the cloud.

​
Skills

Skills are specialized capability packages the Waker can invoke during conversations or triggered tasks.
  • Skill Marketplace — install from the official Qoder Skills Marketplace.
  • Upload a Skill — import custom skill packages from your local machine.
  • Built-in skills — included by default and cannot be uninstalled.
Go to Waker details → Skills to view installed skills, browse the marketplace, or upload your own.

​
Connectors

Connectors are bridges between the Waker and external tools or services. You can add connectors manually.

​
Projects

Projects define the workspace the Waker operates in. Go to Waker details → Projects to:
  • Bind local directories or Git repositories.
  • Add multiple sources to a single project (multiple directories or repositories).
  • Use project-level memory, kept separate from the Waker’s personal memory.

​
Permissions

Permissions control exactly what a Waker can and cannot do. Go to Waker details → Permissions, which covers two main areas plus built-in tool policies. Tool protection validates tool parameters against built-in security rules and triggers an approval request when a high-risk action is detected. Rule categories include:
  • Command injection — detects destructive operations like rm, mv.
  • Resource abuse — detects fork bombs, system restarts, and similar.
  • Code execution — detects remote-execution patterns like curl | bash.
  • Network abuse — detects reverse shells, localhost tunneling.
  • Sensitive file access — detects access to critical system files.
  • Privilege escalation — detects sudo and similar operations.
File protection controls the Waker’s read and write access to local files. Built-in tools lets you set allow / ask / disable policies per built-in tool available to the Waker. When a Waker’s action triggers a security rule, an approval card appears for you to allow or reject it.

​
Example: automated bug fixing

Two ways to drive a bug investigation and fix. Method A — manual conversation. Open the console and select a Waker with the Backend Engineer role. Ask: “Investigate the root cause of this error and propose a minimal fix” and paste in the error details. After you confirm the proposed fix, ask the Waker to apply it and submit a PR. Whenever it needs to run a command or modify a file, it pauses for your approval first. Method B — automated task. Go to Waker details → Triggered Tasks → New. Set the trigger type to Event and configure it to listen for new issues in your GitHub repository. Task description: “When a new bug issue is filed, automatically analyze it and open a fix PR.” Save — every new bug issue triggers analysis and a fix attempt automatically.

​
Feedback

Hit a problem or have a feature request? Click the button in the bottom-right corner of the console to submit feedback — we’ll respond and follow up as soon as possible.