CLI overview

The decimalai command gives you the platform from a terminal. decimalai --help and decimalai <group> --help print what your installed version takes.

Command overview

Group	Commands	What it’s for	Reference
`starter`	`support`	Download a runnable Support project, with no account	Setup
`init`		Check your key, or write a runnable agent file	Setup
`agents`	`create`, `apply-pack`	Create an agent, or apply a role pack to one	Setup
`demo`	`regression`, `skills`, `reset`	Seed guided demos, and remove them	Demo sandbox
`skills`	`search`	Find a skill in the public registry, with no key	Skills commands
`skills`	`add`, `install`, `update`, `remove`, `status`, `verify`, `migrate`, `pull`	Manage registry skills on disk like packages	Skills commands
`skills`	`list`, `sync`, `scan`, `review`, `benchmark`, `push`	Upload, safety-check and measure your own skills	Skills commands
`traces`	`list`, `show`, `stats`, `import`	Read and import runs	Traces
`eval`	`push`	Push scores to a trace	Evaluations
`evaluators`	`add`, `list`, `remove`, `templates`	Manage the checks that score traces	Evaluators
`manifests`	`list`	List an agent’s recorded configurations	Manifests
`regression-check`		Find which traces a change would break	Checks
`compat-check`		Count which traces a manifest change affects	Checks
`repair`	`preview`, `apply`	Fix affected traces without re-running your agent	Repair
`replay`	`run`	Re-run stale traces through your updated agent	Replay
`datasets`	`list`, `show`, `pull`, `export`, `build`, `push-to-hub`	Build training data and get it onto disk	Datasets

Installation

The CLI is included with the Python SDK:

pip install decimalai

Authentication

Every command that talks to your workspace accepts --api-key or the DECIMAL_API_KEY environment variable. skills scan, skills verify and skills remove (offline), skills search and skills pull (public registry), and starter support (public project download) need no DecimalAI key and take no --api-key. skills add, install, update, status --remote and migrate take one, and without it work anonymously, with public skills only:

# Via environment variable (recommended)
export DECIMAL_API_KEY="dai_sk_..."

# Or via flag
decimalai traces list --api-key dai_sk_...

Setup

`decimalai starter support`

Download a new local Support project without saving an agent into a workspace. The browser builder downloads the same project without installing the SDK.

decimalai starter support --project ./support-agent --name my-support-agent
cd support-agent
python3 agent.py --check-setup
export OPENAI_API_KEY="your-provider-key"
python3 agent.py --task 'Draft a useful reply about a duplicate charge.'
python3 agent.py --checks

Option	Purpose
`--project PATH`	Required new output directory; refuses existing directories.
`--name NAME`	Agent name recorded in the frozen snapshot. Default `support-agent`.
`--prompt-file PATH`	Use reviewed local prompt text instead of the pack default.
`--skill-id ID`	Select a public starter identity; repeat for multiple skills. Default is the currently available starter selection.
`--pack-revision HASH`	Require the reviewed revision; fail if the catalog changed.
`--base-url URL`	Public backend URL. No authentication header is sent.

The project requires Python 3.10+ and your OpenAI API access, uses no pip dependencies, and sends no DecimalAI telemetry. It delivers all the selected skills’ instruction text; it doesn’t use the connected SDK router or execute companion scripts. --checks makes two sample-task calls and two judge calls using your provider key. See Skill packs for the hosted trial and connected trace paths.

`decimalai init`

Verify your API key, test connectivity, and send a test trace. It then prints the next commands to try and links to the dashboard and the docs.

$ export DECIMAL_API_KEY="dai_sk_..."
$ decimalai init
  ✓ API key: dai_sk_tes...abc1
  ✓ Connected to workspace: my-team (scope: workspace)
  ✓ Test trace sent successfully

Option	Default	Description
`--no-test-trace`	(sends by default)	Skip sending a test trace

With no key it exits 1, since this command checks a key. Skills need neither a key nor an account, so it starts you there:

  No DecimalAI API key is set, and skills don't need one, or an account:

    decimalai skills search <what your agent does>
    decimalai skills add <name>

  Start here: https://docs.decimal.ai/quickstart

It then says what a free account adds and where to get a key: Settings → API keys.

`decimalai init <agent-name>`

Given the name of an agent that already exists, writes a runnable agent.py wired to it: the agent name bound, its skills loaded at run time, a model on one editable line, and one example call.

$ decimalai init refund-bot

Output:

  ✓ Wrote agent.py — refund-bot, langchain, 3 skills attached, system prompt

  Install:
    pip install "decimalai[langchain]" langchain langchain-openai

  Set:
    export DECIMAL_API_KEY="..." (already set)
    export OPENAI_API_KEY="..."

  Run:
    python agent.py

  The trace appears at:
    → https://app.decimal.ai/agents/refund-bot

The Set: list follows the model: --model anthropic:claude-… on langchain swaps in langchain-anthropic and ANTHROPIC_API_KEY; --framework pydantic-ai defaults the MODEL line to openai:gpt-4o-mini (Pydantic AI requires the provider prefix). decimalai[adk] and decimalai[pydantic-ai] need Python 3.11+. On 3.10, pip installs an older decimalai release for those extras without saying so, so --framework adk and --framework pydantic-ai print a ! line under the install line. The agent’s name is written once, as AGENT_NAME = "refund-bot" near the top, and every template passes it to agent_config(AGENT_NAME), which files each run’s trace under it. To point the file at another agent, change that one line. Every --framework file also runs without DECIMAL_API_KEY: agent_config() then runs keyless. Generated with --base-url, a file sets DECIMAL_BASE_URL before it imports decimalai. The file is yours, and nothing runs on DecimalAI’s side. Skills are fetched at run time, so changing what is attached in the dashboard changes what the file does. Its header lists the skills attached to the agent, and says that it is also offered every skill its workspace owns, unless that skill’s Offered to setting is Only agents it’s assigned to. The agent must exist first: create one at app.decimal.ai/agents/new or with decimalai agents create. An unknown name exits with a suggestion.

Option	Default	Description
`--framework`	`langchain`	`langchain`, `openai-agents`, `pydantic-ai`, `adk` (Gemini only, default `gemini-3.5-flash`), `anthropic` (the Anthropic Messages API with its Tool Runner, default `claude-haiku-5-5`), `openai` (your own Chat Completions tool loop on the OpenAI SDK, default `gpt-4o-mini`) or `google-genai` (your own function-calling loop on the Gemini API, default `gemini-3.5-flash`)
`--out`	`./agent.py`	Where to write
`--project`	none	Create a new Support project directory instead: see Support project with checks. Can’t be combined with `--out` or `--force`
`--model`	framework default	Value for the generated `MODEL` line
`--force`	off	Overwrite an existing file
`--dry-run`	off	Print the file instead of writing it

Only langchain, openai-agents, pydantic-ai, adk, anthropic, openai and google-genai can be scaffolded. Two adapters can deliver skills but have no template yet, crewai and claude-agent-sdk, and are refused with “can deliver skills, but there is no scaffold for it yet”: wire them by hand, as the CrewAI and Claude Agent SDK pages show. llamaindex, autogen and otel are refused with the reason: their adapters have no seam to insert a prompt, so a generated file would trace perfectly and deliver none of the agent’s skills, silently. A file that looks like it works but delivers nothing is worse than no file.

`decimalai agents create`

Creates an agent from a terminal, the same call as the dashboard’s create form. The agent exists as soon as the command returns, with no code deployed yet; decimalai init <agent-name> then writes a runnable file for it.

$ decimalai agents create support-bot --pack support
$ decimalai agents create triage --prompt-file prompt.md --skill my-runbook

Option	Default	Description
`--pack`	none	Start from a role pack: its starter prompt and starter skills, e.g. `support`, `data-engineer`, `doc-ops`
`--no-starters`	off	With `--pack`, take the starter prompt but none of its starter skills
`--skill`	none	Attach a skill: one of yours or a public registry skill, by name, slug or id. Repeatable
`--prompt-file`	none	System prompt from a file (`-` reads stdin). Replaces the pack’s starter prompt
`--description`	none	A note on what this agent does

Skill names resolve exactly, never to the nearest search hit, and an unknown one stops the command before anything is created. A skill the server refuses to attach (a plan limit, say) is listed in the output with the reason, and the agent is still created. A pack starter skill that is unavailable right now is listed with a warning and left off. An existing agent name is refused.

`decimalai agents apply-pack`

Applies a role pack to an agent that already exists, the same change as Start from a pack on the agent’s Prompt & skills page. The pack’s starter prompt becomes the agent’s next prompt version, and its starter skills are added; skills already on the agent stay. It is one setup version on the agent’s Versions page. Your code doesn’t change: agent_config() reads the new prompt and skills on its next run.

$ decimalai agents apply-pack order-support support
$ decimalai agents apply-pack order-support support --dry-run
$ decimalai agents apply-pack triage support --keep-prompt

It shows what will change, asks, then applies:

  Apply the Support pack to order-support:
    Prompt: v1 → v2, the Support starter prompt (15 lines)
    Skills: 2 added, 6 already on the agent
      + sla-breach-response
      + support-reply-format
    Writes setup v2 on the agent's Versions page.

  Apply it? [y/N]: y
  ✓ Applied the Support pack to order-support (setup v2)

Option	Default	Description
`--dry-run`	off	Show what applying would do, and change nothing
`--yes`, `-y`	off	Apply without asking first
`--keep-prompt`	off	Add the pack’s starter skills and leave the agent’s prompt as it is
`--no-starters`	off	Take the pack’s starter prompt but none of its starter skills
`--skill`	none	Also add a skill: one of yours or a public registry skill, by name, slug or id. Repeatable

The new prompt version is labelled Applied the Support pack. Fill in its <blanks> on Prompt & skills; that saves the next version.
Applying the same pack again changes nothing and says Nothing to apply.
A pinned agent keeps its pin. The starter is saved as a new version but not served, and the command says so.
A skill DecimalAI refuses (blocked, or past your plan’s skill limit) is left out and named; the rest applies.
An agent that doesn’t exist is refused, with the agents create --pack command to make it instead. An unknown --skill name stops the command before anything changes.

Support project with checks

Requires the Support pack’s required starters, LangChain, and setup checks enabled on the platform.

Start from the Support pack, review the selected skills and create a draft-response agent. Have Python 3.10+, pip and a funded OpenAI API key ready. Use the name and DecimalAI key shown on your agent page.

python -m venv .venv
source .venv/bin/activate
python -m pip install "decimalai[langchain,evals]" langchain langchain-openai
export DECIMAL_API_KEY="your-decimalai-key"
export OPENAI_API_KEY="your-openai-key"
decimalai init support-drafter --project support-drafter-project
cd support-drafter-project
python -m pip install -r requirements.txt
python -m pip freeze > requirements.lock.txt
python check_agent.py

--project refuses to overwrite an existing project. It writes agent.py, check_agent.py, checks.json, project.json, requirements.txt, .env.example, .gitignore and README.md. The default model is gpt-5.4-2026-03-05. A model judge also uses OpenAI, so checks incur provider charges. In a fresh deployment environment, install requirements.lock.txt and rerun the checks there. A pass requires the matching traces to arrive with the expected delivered configuration. The Support starter project says what a pass does and doesn’t mean. If trace confirmation is delayed, use python check_agent.py --resume, with --check-id <id> for a specific attempt. This sends no new model requests. Keep the per-attempt check-results.<id>.json receipt; configuration edits make prior evidence stale and require another check. The generated agent drafts replies only and has no tools, so review replies before sending them. Replace the sample policies with your own, and follow the generated README to deploy.

Demo sandbox

One-command guided demos that seed realistic data into your workspace, so you can see the headline workflows before instrumenting anything. Demo names start with [Demo] , and demo reset removes them.

Option (regression / skills)	Default	Description
`--reset / --no-reset`	`--reset`	Wipe existing demo data before seeding, so you always land in a clean state
`--web`	(derived from `--base-url`)	Frontend base URL for the printed links

`decimalai demo regression`

Seed the “Your agent changed” demo: a v1→v2 agent (model swap, tool rename/removal, prompt rewrite) plus a trace corpus, then run the regression check and print a deep link to the impact report.

$ decimalai demo regression

`decimalai demo skills`

Seed the “Find skills that work” demo: three workspace-scoped skills with deliberately varied effectiveness, plus the stats recompute, so the registry ranks them by seeded data. Nothing lands in the public registry.

$ decimalai demo skills

`decimalai demo reset`

Remove all demo data for the workspace (both demos). It matches the demos’ exact names, so an agent or skill of yours whose name merely starts with [Demo] is left alone.

$ decimalai demo reset

Global options

All commands that talk to your workspace accept these options:

Option	Default	Description
`--api-key`	`$DECIMAL_API_KEY`	API key
`--base-url`	`$DECIMAL_BASE_URL`, else `https://api.decimal.ai`	Platform URL

The exceptions are in Authentication: of those, skills pull and starter support still accept --base-url. The skills-on-disk commands also take --directory DIR, the project root.

Environment variables

Variable	Description
`DECIMAL_API_KEY`	API key (used by every authenticated command)
`DECIMALAI_API_KEY`	Alias (also accepted)
`DECIMAL_BASE_URL`	Platform API URL
`DECIMALAI_BASE_URL`	Alias (also accepted)
`DECIMAL_AGENT_NAME`	Default agent name for `decimalai.init()` and every call that is given none: `instrument()`, `start_trace()`, `@decimalai.trace` and `SkillRouter`. A name passed to `init()` or to the call wins.
`DECIMAL_AUTO_TRACE`	Auto-instrument on import. Accepts any framework or provider flag name: `langchain`, `openai-agents`, `llamaindex`, `crewai`, `otel`, or a direct provider SDK (`openai`, `anthropic`, `google`). `autogen` is accepted and maps to `otel`: AutoGen / AG2 is not an integration. On `crewai` it also turns the skill loader on, as `init(crewai=True)` does; `langchain` and `openai-agents` trace only. `adk` traces nothing and logs a warning: put `DecimalAIPlugin()` in your ADK `App`.
`DECIMALAI_SKILLS`	Whether `init(crewai=True)`, and so `DECIMAL_AUTO_TRACE=crewai`, delivers skills. It is the default of `init(skills=...)`. `0`, `false`, `no` or `off` keeps it to tracing only, and on `langchain` silences the warning a run without `agent_config()`’s middleware logs; unset, empty, `1`, `true`, `yes` or `on` delivers skills; any other value is treated as off and logs a warning. An explicit `skills=` argument wins, and an explicit `instrument(enable_skill_loader=True)` is not affected.

Python SDK

Skills in your repo

CLI

REST

Command overview

Installation

Authentication

Setup

`decimalai starter support`

`decimalai init`

`decimalai init <agent-name>`

`decimalai agents create`

`decimalai agents apply-pack`

Support project with checks

Demo sandbox

`decimalai demo regression`

`decimalai demo skills`

`decimalai demo reset`

Global options

Environment variables

​Command overview

​Installation

​Authentication

​Setup

​decimalai starter support

​decimalai init

​decimalai init <agent-name>

​decimalai agents create

​decimalai agents apply-pack

​Support project with checks

​Demo sandbox

​decimalai demo regression

​decimalai demo skills

​decimalai demo reset

​Global options

​Environment variables

Command overview

Installation

Authentication

Setup

`decimalai starter support`

`decimalai init`

`decimalai init <agent-name>`

`decimalai agents create`

`decimalai agents apply-pack`

Support project with checks

Demo sandbox

`decimalai demo regression`

`decimalai demo skills`

`decimalai demo reset`

Global options

Environment variables