Commands Reference

This page documents all available Chamber CLI commands and their options.

Global Flags

These flags are available on all commands:

Flag	Description
`-o, --output`	Output format: `table` (default), `json`, `yaml`
`-h, --help`	Show help for any command

AI Chat

`chamber chat`

Ask questions about your GPU infrastructure using natural language. The assistant can also suggest and execute CLI commands on your behalf with confirmation. See AI Assistant for the complete guide.

# Quick question
chamber chat "what jobs are running right now?"

# Interactive conversation
chamber chat

# Named conversation (resume later)
chamber chat -c my-thread "help me debug this failure"

# Pipe input for analysis
kubectl logs my-pod | chamber chat "what errors are in these logs?"

# Attach an image
chamber chat -i gpu-metrics.png "what does this graph show?"

# List recent conversations
chamber chat --list

Flag	Description
`-c, --conversation`	Named conversation thread (alphanumeric, hyphens, underscores, max 64 chars)
`-m, --model`	AI model selection
`-i, --image`	Attach image file (PNG, JPG, GIF, WebP) — single-shot and piped modes only
`--no-stream`	Buffer full response (auto-enabled when stdout is piped)
`--raw`	Disable markdown rendering
`-l, --list`	List recent conversations

Chat requires browser-based login (chamber login). API tokens (--token) are not supported for chat.

Authentication

Authenticate with Chamber.

# Browser-based login (interactive)
chamber login

# Token-based login (headless/scripted)
chamber login --token ch.your-api-token

Flag	Description
`-t, --token`	API token for authentication

`chamber logout`

Remove stored credentials.

chamber logout

`chamber whoami`

Display current user information.

chamber whoami

Example output:

Email: user@example.com
User ID: 550e8400-e29b-41d4-a716-446655440000
Token expires: 2024-01-15T10:30:00Z
Organization: my-org-id

Run (Auto-Containerize)

The run command automatically containerizes and submits your training code. See Auto-Containerize with Run for complete documentation.

`chamber run`

Auto-containerize and submit a GPU workload.

# Preview what will be generated
chamber run ./my-project --gpus 4 --gpu-type H100 --team <team-id> --dry-run

# Build, push, and submit
chamber run ./my-project --gpus 4 --gpu-type H100 --team <team-id>

# With custom entrypoint
chamber run ./my-project \
  --gpus 8 \
  --gpu-type H100 \
  --team <team-id> \
  --entrypoint train.py \
  --entrypoint-args "--config config.yaml"

# Use reserved capacity
chamber run ./my-project --gpus 4 --gpu-type H100 --team <team-id> --class RESERVED

Flag	Description
`--team, -t`	Team ID (required)
`--gpus, -g`	Number of GPUs
`--gpu-type`	GPU type: `H100`, `A100`, `L40S`, etc.
`--class, -c`	Workload class: `RESERVED` or `ELASTIC`
`--name, -n`	Workload name
`--entrypoint`	Override detected Python entrypoint
`--entrypoint-args`	Arguments for the entrypoint
`--registry`	Container registry URL
`--dockerfile`	Use existing Dockerfile
`--dry-run`	Preview without building/submitting
`--save-dockerfile`	Save generated Dockerfile
`--save-manifest`	Save generated K8s manifest
`--no-cache`	Force Docker rebuild

Use --dry-run to preview the generated Dockerfile and manifest before submitting.

Prerequisites: Docker, AWS CLI (for ECR). See Auto-Containerize with Run for setup instructions.

Workloads

Manage GPU workloads. Aliases: wl, jobs

`chamber workloads list`

List workloads with optional filters.

# List all workloads
chamber workloads list

# Filter by status
chamber workloads list --status RUNNING

# Filter by initiative
chamber workloads list --initiative abc123

# Limit results
chamber workloads list --limit 50

# Output as JSON
chamber workloads list -o json

Flag	Description
`-s, --status`	Filter by status: `PENDING`, `QUEUED`, `RUNNING`, `COMPLETED`, `FAILED`, `CANCELLED`, `PREEMPTED`
`-i, --initiative`	Filter by team ID
`--submitted-by`	Filter by submitter user ID
`-l, --limit`	Maximum results (1-100, default: 20)

Example output:

ID        Name                    Status    GPUs  Type  Class     Submitted
--------  ----------------------  --------  ----  ----  --------  ----------------
a1b2c3d4  gpt-finetune-v2         RUNNING   8     H100  RESERVED  2024-01-15T09:30
e5f6g7h8  embedding-training      QUEUED    4     A100  ELASTIC   2024-01-15T09:15

`chamber workloads get`

Get detailed information about a specific workload.

chamber workloads get <workload-id>

# Output as JSON
chamber workloads get <workload-id> -o json

Example output:

ID:            a1b2c3d4-e5f6-7890-abcd-ef1234567890
Name:          gpt-finetune-v2
Status:        RUNNING
Workload Class: RESERVED
Priority:      75
GPUs:          8 x H100
Initiative:    ml-training-team
Submitted By:  user@example.com
Submitted At:  2024-01-15T09:30:00Z
Started At:    2024-01-15T09:32:15Z

`chamber workloads stats`

Get aggregated workload statistics.

# Default: last 7 days
chamber workloads stats

# Custom time range
chamber workloads stats --time-range last_30_days

# Group by submitter
chamber workloads stats --group-by submitted_by

Flag	Description
`--time-range`	Time range: `today`, `this_week`, `this_month`, `last_7_days` (default), `last_30_days`
`--group-by`	Group results by: `submitted_by`, `initiative_id`, `instance_type`

`chamber workloads metrics`

Get GPU metrics for a running workload.

chamber workloads metrics <workload-id>

Example output:

GPU Metrics:

Metric           Current  Avg   Min   Max
---------------  -------  ----  ----  ----
Utilization (%)  92.5     88.3  45.0  98.0
Memory (%)       78.2     75.1  60.0  82.0
Temperature (C)  72       68    55    76
Power (W)        420      385   250   450

`chamber workloads cancel`

Cancel a running or queued workload.

chamber workloads cancel <workload-id>

Submit Workloads

`chamber submit`

Submit a new GPU workload.

# Interactive mode
chamber submit -i

# With flags
chamber submit \
  --name "my-training-job" \
  --gpus 8 \
  --gpu-type H100 \
  --class RESERVED \
  --initiative abc123

# With Kubernetes manifest
chamber submit \
  --name "custom-job" \
  --gpus 4 \
  --gpu-type A100 \
  --class ELASTIC \
  --initiative abc123 \
  --manifest job.yaml

# Using a template
chamber submit \
  --name "my-job" \
  --initiative abc123 \
  --template template-id

Flag	Description
`-n, --name`	Workload name (required)
`--initiative`	Team ID (required)
`-g, --gpus`	Number of GPUs
`--gpu-type`	GPU type: `H100`, `A100`, etc. (required unless using template)
`-c, --class`	Workload class: `RESERVED` or `ELASTIC` (default: `ELASTIC`)
`-p, --priority`	Priority: `critical`, `high`, `normal`, `low`, `background`
`--external-id`	Your own workload identifier
`-m, --manifest`	Path to Kubernetes manifest YAML
`--template`	Use a template instead of specifying flags
`--scaling-mode`	`gang` or `elastic` (default: `gang`)
`--pods`	Number of pods for distributed training
`--gpus-per-pod`	GPUs per pod (manual distribution)
`--min-pods`	Minimum pods for elastic scaling
`-i, --interactive`	Interactive mode with prompts

Use --class RESERVED for guaranteed capacity that won’t be preempted. Use --class ELASTIC for lower-priority workloads that can run on spare capacity.

Capacity

`chamber capacity`

View capacity pools and budget information.

chamber capacity

# Output as JSON
chamber capacity -o json

Example output:

Budget:
  Allocated: 10000 GPU-hours
  Used:      4500 GPU-hours
  Available: 5500 GPU-hours

Capacity Pools:
ID        Name              GPU Type  Total  Used  Available
--------  ----------------  --------  -----  ----  ---------
pool-001  us-west-h100      H100      64     48    16
pool-002  us-east-a100      A100      128    96    32

Teams

Manage teams. Aliases: team, initiatives

`chamber teams list`

List all teams you have access to.

# List all teams
chamber teams list

# Output as JSON
chamber teams list -o json

Example output:

ID        Name                 Description                  Parent
--------  -------------------  ---------------------------  --------
team-001  ML Research          Main ML research team        -
team-002  NLP Projects         Natural language processing  team-001
team-003  Computer Vision      CV and image processing      team-001

`chamber teams get`

Get detailed information about a specific team.

chamber teams get <team-id>

`chamber teams create`

Create a new team or sub-team.

# Create a root team (requires ORG_ADMIN)
chamber teams create --name "New Team" --description "Team description"

# Create a sub-team under an existing team
chamber teams create --name "Sub Team" --parent parent-team-id

Flag	Description
`--name`	Team name (required)
`--description`	Team description
`--parent`	Parent team ID (for creating sub-teams)

Creating root-level teams requires ORG_ADMIN permissions. Sub-teams can be created by team members.

Allocations

Manage capacity allocations for teams. Aliases: alloc

`chamber allocations list`

List allocations for a team.

chamber allocations list --initiative <team-id>

Flag	Description
`--initiative`	Team ID (required)

Example output:

ID          Reservation    Instances  Status   Expires
----------  -------------  ---------  -------  ----------------
alloc-001   res-h100-west  4          ACTIVE   2024-06-30T23:59
alloc-002   res-a100-east  8          ACTIVE   2024-12-31T23:59

`chamber allocations get`

Get detailed information about a specific allocation.

chamber allocations get <allocation-id>

`chamber allocations create`

Create a new allocation for a team. Requires ORG_ADMIN permissions.

chamber allocations create \
  --initiative <team-id> \
  --reservation <reservation-id> \
  --instances 4

Flag	Description
`--initiative`	Team ID (required)
`--reservation`	Reservation ID (required)
`--instances`	Number of instances to allocate (required)

Templates

Manage reusable workload templates. Aliases: template, tpl

`chamber templates list`

List available templates.

# List all templates
chamber templates list

# Filter by name
chamber templates list --name "training"

# Filter by scope
chamber templates list --scope ORGANIZATION

# Include system templates
chamber templates list --include-system

Flag	Description
`--name`	Filter by template name
`--scope`	Filter by scope: `PERSONAL`, `PROJECT`, `ORGANIZATION`
`--include-system`	Include system-provided templates

Example output:

ID           Name                  Scope         GPU Type  GPUs
-----------  --------------------  ------------  --------  ----
tpl-001      standard-training     ORGANIZATION  H100      8
tpl-002      inference-job         ORGANIZATION  A100      4
tpl-003      my-custom-template    PERSONAL      H100      16

`chamber templates get`

Get detailed information about a specific template.

chamber templates get <template-id>

Artifacts (Container Images)

Manage container images and artifact repositories. Aliases: artifact, images Chamber supports multiple registry types:

Doogle - Doogle Artifact Repository (default, managed by Chamber)
ECR - AWS Elastic Container Registry
GCR - Google Container Registry
ACR - Azure Container Registry

`chamber artifacts repos list`

List artifact repositories in your organization.

# List all repositories
chamber artifacts repos list

# Filter by registry type
chamber artifacts repos list --type doogle

# Output as JSON
chamber artifacts repos list -o json

Flag	Description
`--type`	Filter by registry type: `doogle`, `ecr`, `gcr`, `acr`
`-l, --limit`	Maximum results (default: 20)

Example output:

ID        Name              Type    URI                                    Images  Created
--------  ----------------  ------  -------------------------------------  ------  ----------------
repo-001  ml-training       doogle  doogle.usechamber.io/org123/ml-train   12      2024-01-15T09:30
repo-002  inference-models  ecr     123456.dkr.ecr.us-east-1.amazonaws.com 5       2024-01-10T14:20

`chamber artifacts repos get`

Get detailed information about a specific repository.

chamber artifacts repos get <repository-id>
chamber artifacts repos get <repository-id> -o json

`chamber artifacts repos create`

Create a new artifact repository.

# Create a Doogle repository (recommended)
chamber artifacts repos create --name my-models

# Create with description
chamber artifacts repos create --name training-images --description "Training container images"

# Create ECR repository
chamber artifacts repos create --name my-ecr-repo --type ecr

Flag	Description
`-n, --name`	Repository name (required)
`-d, --description`	Repository description
`--type`	Registry type: `doogle` (default), `ecr`, `gcr`, `acr`

Use Doogle Artifact Repository for a fully managed experience with no additional cloud setup required.

`chamber artifacts repos delete`

Delete an artifact repository.

chamber artifacts repos delete <repository-id>

This will delete all images in the repository. This action cannot be undone.

`chamber artifacts list`

List container images in a repository.

# List images in a repository
chamber artifacts list --repo <repository-id>

# Filter by tag
chamber artifacts list --repo <repository-id> --tag latest

# Output as JSON
chamber artifacts list --repo <repository-id> -o json

Flag	Description
`-r, --repo`	Repository ID (required)
`-t, --tag`	Filter by image tag
`-l, --limit`	Maximum results (default: 20)

Example output:

Digest                Tags          Size      Pushed At
--------------------  ------------  --------  ----------------
sha256:a1b2c3d4...    latest, v1.2  1.2 GB    2024-01-15T10:30
sha256:e5f6g7h8...    v1.1          1.1 GB    2024-01-14T09:15
sha256:i9j0k1l2...    <untagged>    980 MB    2024-01-13T14:45

`chamber artifacts get`

Get detailed information about a specific artifact (image).

chamber artifacts get <artifact-id>
chamber artifacts get <artifact-id> -o json

`chamber artifacts delete`

Delete an artifact (image) from a repository.

chamber artifacts delete <artifact-id>

Get Docker login credentials for pushing/pulling images.

# Get credentials
chamber artifacts login

# Get credentials for a specific repository
chamber artifacts login <repository-id>

# Output docker login command (can be piped to sh)
chamber artifacts login --docker

# Login directly with Docker
chamber artifacts login --docker | sh

Flag	Description
`--docker`	Output full docker login command

Example output:

Registry:     doogle.usechamber.io
Username:     chamber
Password:     eyJhbGc...
Expires At:   2024-01-15T16:30:00Z

To login with Docker:
  docker login -u chamber -p <password> doogle.usechamber.io

Metrics

`chamber metrics`

View organization-wide GPU metrics.

# Default: last 24 hours
chamber metrics

# Filter by team
chamber metrics --initiative <team-id>

# Custom time range
chamber metrics --time-range last_7d

# Group by team
chamber metrics --group-by team

# Output as JSON
chamber metrics -o json

Flag	Description
`--initiative`	Filter by team ID
`--time-range`	Time range: `last_1h`, `last_6h`, `last_24h` (default), `last_7d`, `last_30d`
`--group-by`	Group metrics by: `team`

Configuration

`chamber config show`

Display current configuration.

chamber config show

`chamber config get`

Get a specific configuration value.

chamber config get organization_id
chamber config get api_url

`chamber config set`

Set a configuration value.

chamber config set organization_id your-org-id
chamber config set api_url https://api.usechamber.io/v1

Key	Description
`organization_id`	Default organization for API requests
`api_url`	Chamber API URL

System

`chamber health`

Check API health status.

chamber health

`chamber completion`

Generate shell completion scripts.

chamber completion bash
chamber completion zsh
chamber completion fish

JSON Output

All commands support -o json for machine-readable output:

# Get workloads as JSON
chamber workloads list -o json

# Pipe to jq for filtering
chamber workloads list -o json | jq '.[] | select(.status == "RUNNING")'

# Use in scripts
WORKLOAD_ID=$(chamber workloads list -o json | jq -r '.[0].id')

Early Access — The Chamber CLI is currently in early access. Contact support@usechamber.io to get started.

​Commands Reference

​Global Flags

​AI Chat

​chamber chat

​Authentication

​chamber login

​chamber logout

​chamber whoami

​Run (Auto-Containerize)

​chamber run

​Workloads

​chamber workloads list

​chamber workloads get

​chamber workloads stats

​chamber workloads metrics

​chamber workloads cancel

​Submit Workloads

​chamber submit

​Capacity

​chamber capacity

​Teams

​chamber teams list

​chamber teams get

​chamber teams create

​Allocations

​chamber allocations list

​chamber allocations get

​chamber allocations create

​Templates

​chamber templates list

​chamber templates get

​Artifacts (Container Images)

​chamber artifacts repos list

​chamber artifacts repos get

​chamber artifacts repos create

​chamber artifacts repos delete

​chamber artifacts list

​chamber artifacts get

​chamber artifacts delete

​chamber artifacts login

​Metrics

​chamber metrics

​Configuration

​chamber config show

​chamber config get

​chamber config set

​System

​chamber health

​chamber completion

​JSON Output

Commands Reference

Global Flags

AI Chat

`chamber chat`

Authentication

`chamber login`

`chamber logout`

`chamber whoami`

Run (Auto-Containerize)

`chamber run`

Workloads

`chamber workloads list`

`chamber workloads get`

`chamber workloads stats`

`chamber workloads metrics`

`chamber workloads cancel`

Submit Workloads

`chamber submit`

Capacity

`chamber capacity`

Teams

`chamber teams list`

`chamber teams get`

`chamber teams create`

Allocations

`chamber allocations list`

`chamber allocations get`

`chamber allocations create`

Templates

`chamber templates list`

`chamber templates get`

Artifacts (Container Images)

`chamber artifacts repos list`

`chamber artifacts repos get`

`chamber artifacts repos create`

`chamber artifacts repos delete`

`chamber artifacts list`

`chamber artifacts get`

`chamber artifacts delete`

`chamber artifacts login`

Metrics

`chamber metrics`

Configuration

`chamber config show`

`chamber config get`

`chamber config set`

System

`chamber health`

`chamber completion`

JSON Output