HENGSHI CLI

Overview

HENGSHI CLI is the official command-line tool from HENGSHI and the unified command entrypoint for both humans and AI agents. It turns actions that used to live in UI clicks, ad-hoc scripts, or private API calls into stable, reviewable, and automatable command workflows.

Its current coverage includes:

• Data connections and dataset onboarding
• HQL / HE queries and data preview
• Dashboard, report, and element creation / configuration
• Permission lookup, grant, and revoke flows
• System configuration, environment checks, and operational tasks
• Agent execution workflows powered by official skills

NOTE

Starting with 2.0.0, the public command name is hbi, and this documentation uses hbi plus HBI_* as the primary namespace. To avoid confusion, this page does not document legacy naming/compatibility details; follow the behavior of your installed package and upgrade scripts if you still have older names in your environment.

The bridge between HENGSHI SENSE and agents

HENGSHI CLI does not replace the HENGSHI SENSE UI. Its role is to turn core HENGSHI SENSE capabilities into a stable command interface that people and agents can call, review, and reuse. It works for direct terminal users as well as OpenClaw, CI jobs, and automation scripts.

In that chain, the common split looks like this:

Layer	Role	Typical responsibility
Human	Task owner and reviewer	Provide intent, approve boundaries, review outcomes
Agent / automation flow	Orchestration layer	Break work into steps, sequence actions, and decide which command to call next
HENGSHI CLI + skills	Bridge and execution layer	Translate HENGSHI SENSE capabilities into structured commands, outputs, and --dry-run previews
HENGSHI SENSE backend	Capability layer	Actually perform connection, query, modeling, authorization, and delivery actions

If you are designing reusable workflows for OpenClaw or other agents, CLI is usually the bridge itself: it connects agent-side orchestration with real HENGSHI SENSE capabilities instead of forcing the agent to guess UI paths or assemble private APIs on its own.

What problems CLI is good at solving

Scenario	What CLI brings
Letting an agent perform BI delivery tasks	One command model and one output contract instead of “every agent invents its own API calls”
Putting risky changes into review flows	--dry-run lets you inspect changes before execution
Returning results to humans or downstream systems	json / yaml / table outputs are easier to read, record, and post-process
Building repeatable operational workflows	Local terminals, CI, background services, and agent runtimes can all reuse the same commands
Surfacing out-of-browser changes back into the UI	SSE and the Autopilot indicator provide a realtime feedback bridge

Command surface at a glance

HENGSHI CLI is not a single-purpose tool. It is a command tree aligned with the main BI delivery chain.

Command domain	What it covers	Example
app	Applications, spaces, portals, shared dimensions	hbi app list --area personal-area --root
dataset	Dataset onboarding and listing	hbi dataset list --app 4812 --output json
data-model	HQL / HE querying and data preview	hbi data-model query --app 4812 --dataset 3 "SUM({amount})"
dashboard	Dashboards, reports, and themes	hbi dashboard create --app 4812 "Regional Sales Cockpit"
element	Charts, filters, containers, buttons, iframes, and more; supports custom titles and content recognition (link/email/image) flags	hbi element chart create --dashboard 2048 --app 4812 --dataset 3 table --identify-link true --identify-email true --identify-image true
connection	Data source connection management	hbi connection list
authorize	Unified authorization API	hbi authorize grant app 42 --user 123:editor --dry-run
preferences	System configuration inspection and previewed updates	hbi preferences everest --output json

HQL expression support

The --dimension, --measure, and --where flags accept HQL (HENGSHI Query Language) expressions, not just plain field names.

• Time bucketing: Prefer calendar HQL functions (e.g., day(), month(), trunc_month()) over string-slicing patterns like substring / date_format
• Cross-period functions: HQL also provides moy(), dom(), dow(), doy(), woy(), qoy() for cross-period time aggregation — semantically equivalent to the UI options like "Month (cross-year)" and "Day (cross-month)"
• Expression validation: Use hbi dataset expression-rewrite to validate and rewrite expressions

See element chart create --dimension for details.

Official skills and agent execution

CLI is delivered together with official bundled skills. Official skills are typically organized by capability domain; when documenting them, prefer showing both the skill names and the responsibility summary:

• hbi-core: authentication, configuration, output, and terminology rules
• hbi-data, hbi-data-modeling, hbi-pipeline, hbi-notebook: connections, datasets, modeling, execution, and metric workflows
• hbi-dashboard, hbi-dashboard-taste, hbi-app: dashboard planning, page layout, element configuration, and app surfaces
• hbi-permission, hbi-user-mgmt: access lookup, grant/revoke flows, and user/organization governance
• hbi-workflow: cross-domain orchestration and sequencing control

The purpose of these skills is not to replace CLI. They encode execution experience such as which state should be read first, which actions must go through --dry-run, and how results should be returned to people or other systems.

Recommended reading path

If you are new to HENGSHI CLI, read in this order:

1. Installation & Upgrade
2. Authentication & Connection
3. Agent Integration & Skills
4. Common Workflows
5. Command Reference
6. Realtime Sync & Autopilot
7. Troubleshooting & FAQ

Recommendation

If your goal is to let an agent execute stable actions instead of typing commands manually, start with Agent Integration & Skills and Common Workflows.