| title | CLI Reference | ||||
|---|---|---|---|---|---|
| description | Complete reference for all Qualixar OS CLI commands — task execution, team design, memory, config, server, and advanced dispatch | ||||
| category | cli | ||||
| tags |
|
||||
| last_updated | 2026-04-13 |
Qualixar OS ships a qos CLI binary built with Commander.js. It exposes 25 commands across 7 groups, plus access to the full 25-command Universal Command Protocol (UCP) via qos cmd and qos dispatch.
Config is loaded from ~/.qualixar-os/config.yaml, with env vars from ~/.qualixar-os/.env.
Run a task through the full orchestrator pipeline (Forge team design, agent execution, judge evaluation).
qos run <prompt> [options]| Option | Description | Default |
|---|---|---|
-t, --type <type> |
Task type: code, research, analysis, creative, custom |
custom |
-m, --mode <mode> |
Execution mode: companion, power |
companion |
-b, --budget <usd> |
Budget limit in USD | none |
--topology <name> |
Force a specific swarm topology | auto-selected |
--simulate |
Dry-run simulation before execution | off |
--stream |
Stream output in real-time | off |
--template <name> |
Use a named template | none |
qos run "Review auth module for security vulnerabilities" -t code -m power
qos run "Analyze competitor pricing" -t research --budget 0.50
qos run "Build a REST API" --topology pipeline --streamShow the current status of a task.
qos status <taskId>Display the output and artifacts of a completed or failed task.
qos output <taskId>Pause a running task. Agents stop at their next checkpoint.
qos pause <taskId>Resume a previously paused task.
qos resume <taskId>Cancel a task permanently.
qos cancel <taskId>Show cost summary (token usage, USD spent) for a specific task or globally.
qos cost [taskId]Show the Forge design library. Forge stores reusable team designs (topology, agents, tools) indexed by task type.
qos forge [taskType]qos forge # list all designs
qos forge code # list designs for code tasksList registered agents, or show agents assigned to a specific task.
qos agents [taskId]qos agents # list all registered agents
qos agents abc-123-def # show agents for a specific taskShow judge evaluation results for a task, or all recent results.
qos judges [taskId]Output includes judge model, verdict, and score for each evaluation.
Search the SLM-Lite memory system. Returns matching entries with layer and content.
qos memory <query> [options]| Option | Description | Default |
|---|---|---|
-l, --layer <layer> |
Filter by memory layer | all layers |
--limit <n> |
Maximum results | 10 |
qos memory "authentication patterns" --limit 5
qos memory "API design" -l decisionsRead or write configuration. Values are persisted to ~/.qualixar-os/config.yaml.
qos config # show full config
qos config <key> # read a value (dot notation)
qos config <key> <value> # set a value (dot notation)qos config models.primary # read primary model
qos config models.primary claude-sonnet-4-6 # set primary model
qos config mode power # switch to power modeInteractive setup wizard. Creates ~/.qualixar-os/ with config, env, and optionally runs a first task.
qos init [options]| Option | Description |
|---|---|
--no-interactive |
Skip prompts, use defaults + flags |
--default |
Alias for --no-interactive |
--provider <name> |
Primary LLM provider |
--api-key-env <var> |
Env var name for API key |
--model <name> |
Primary model name |
--channels <list> |
Comma-separated channels (e.g. cli,mcp,http) |
--budget <usd> |
Budget limit in USD |
--skip-first-task |
Skip the post-install demo task |
--dashboard-port <port> |
Dashboard port number |
qos init
qos init --default --provider azure --model gpt-4o --budget 5List all models in the catalog with provider, quality score, and max tokens.
qos modelsRun a health check. Validates config, checks connectivity, verifies dependencies.
qos doctorStart the HTTP/WebSocket/A2A server. This is the runtime that plugins and the dashboard connect to.
qos serve [options]| Option | Description | Default |
|---|---|---|
-p, --port <port> |
Port number | 3000 |
--dashboard |
Also serve the dashboard UI | off |
qos serve --dashboard -p 3000Start a dedicated dashboard server.
qos dashboard [options]| Option | Description | Default |
|---|---|---|
-p, --port <port> |
Port number | 3333 |
Start the MCP server on stdio transport. This is what Claude Code connects to for tool access.
qos mcpTypically invoked automatically by Claude Code via the plugin config (npx qualixar-os --mcp), not run manually.
Execute any Universal Command Protocol (UCP) command. The UCP defines 25 commands across 9 categories (task, context, workspace, agents, forge, quality, memory, system, interop) that work identically across all transports.
qos cmd <command> [options]| Option | Description |
|---|---|
--json |
Output raw JSON |
-i, --input <json> |
JSON input for the command |
qos cmd run -i '{"prompt":"analyze code","type":"code"}'
qos cmd agents.list --json
qos cmd forge.design -i '{"taskType":"research"}'
qos cmd memory.search -i '{"query":"auth patterns"}'
qos cmd config.get -i '{"key":"mode"}'List all 25 UCP commands with their category and description.
qos cmd-listLower-level UCP dispatcher. Same as cmd but always outputs JSON.
qos dispatch <command> [jsonInput]qos dispatch context.add '{"paths":["./src"]}'
qos dispatch forge.topologiesImport an agent definition from an external format (SOUL.md, OpenClaw, DeerFlow, NemoClaw, GitAgent).
qos import <path> [options]| Option | Description |
|---|---|
-f, --format <format> |
Source format (auto-detected if omitted) |
qos import ./agents/reviewer/SOUL.md
qos import ./agent.json -f openclawExport an agent to an external format.
qos export <agentId> [options]| Option | Description | Default |
|---|---|---|
-f, --format <format> |
Target format: soul-md, json, yaml |
json |
-o, --output <path> |
Write to file instead of stdout | stdout |
qos export abc-123 -f soul-md -o ./SOUL.md
qos export abc-123 -f yamlScaffold a new Qualixar OS project from a template.
qos new <project>qos new my-agent-pipelineShow the installed Qualixar OS version.
qos versionAll 25 Universal Commands accessible through qos cmd <name>:
| Command | Category | Description |
|---|---|---|
run |
task | Run a new task through the orchestrator pipeline |
status |
task | Get the current status of a running task |
output |
task | Retrieve the output of a completed task |
cancel |
task | Cancel a running task |
pause |
task | Pause a running task |
resume |
task | Resume a paused task |
steer |
task | Redirect a running task with a new prompt |
list |
task | List tasks with optional status filter and pagination |
context.add |
context | Add file paths and URLs as context for a task |
context.scan |
context | Scan a directory for context files |
context.list |
context | List all context entries for a task |
workspace.set |
workspace | Set the working directory for tasks |
workspace.files |
workspace | List generated files from a task output directory |
agents.list |
agents | List agents, optionally filtered by task or status |
agents.inspect |
agents | Inspect a single agent with full details and call history |
forge.design |
forge | Design a team for a task type (dry-run if no taskId) |
forge.topologies |
forge | List all available swarm topologies |
judges.results |
quality | Retrieve judge assessment results for a task |
memory.search |
memory | Search memory entries by query with optional layer filter |
memory.store |
memory | Store a new memory entry in a specific layer |
config.get |
system | Get runtime configuration by dot-notated key path |
config.set |
system | Set a runtime configuration value |
models.list |
system | List all known LLM models with pricing and quality scores |
cost.summary |
system | Get cost summary for a specific task or globally |
import |
interop | Import an agent definition from an external format |