Skip to content

Erupt AI Claw ​

INFO

Erupt AI Claw lets you drive the server directly through natural language — operating annotation-driven data and business, executing shell commands, reading and writing files, and extending custom Skills, as simply as chatting with a colleague.

A self-evolving agent that works and learns simultaneously, supports dozens of models interchangeably, auto-expands its skill library, and intelligently searches conversation memory. The agent learns while working and can run on an ordinary computer.

Erupt Engine

Usage ​

xml
<!-- Supported from version 1.14.1 and above -->
<dependency>
    <groupId>xyz.erupt</groupId>
    <artifactId>erupt-ai-claw</artifactId>
    <version>${LATEST}</version>
</dependency>
yaml
erupt:
  ai:
    claw:
      # Whether to enable Claw
      enabled: true
      # Whether to allow executing shell commands (default true; recommended false in production)
      enable-exec-shell: true  # v1.14.3+

WARNING

Claw has full tool permissions by default. In production environments, restrict the available Tool scope via role configuration (supported from v1.14.3+, see Role-Level Tool Authorization). Security first!

Token Preparation ​

The effectiveness of Claw varies by model. Mainstream public models are recommended.

Claw and SKILLS ​

INFO

Supports interaction with 700,000+ Skills, and can also dynamically create Skills via prompts. Erupt Claw will automatically match and execute the corresponding Skill based on the prompt.

Skill marketplace: https://skills.sh

shell
# You can also manually place skill files in the following directory
~/.erupt/skills/

Creating a SKILL

Executing a SKILL

Claw and Erupt Model Interaction ​

Creating an organizational structure via conversation (data insertion):

Data Update

Data Creation and Content Generation

Querying Model Data

Claw and Operating System Interaction ​

Claw and Browser Interaction ​

This can be achieved indirectly via the MCP feature. Add the following configuration in the MCP menu:

json
{
  "args": [
    "-y",
    "chrome-devtools-mcp@latest"
  ],
  "command": "npx"
}

Claw and Long-Term Memory ​

erupt-ai-claw supports cross-session long-term memory persistence via the Memory tool.

During a conversation, the AI automatically judges and writes important information (technical decisions, user preferences, project context, etc.) to memory files. At the start of the next session, relevant memories are automatically loaded — no need to re-explain the background.

Memory files are stored by default at:

shell
# You can also manually create .md files in this directory; the AI will automatically recognize and read them in the next session.
~/.erupt/memory/

Claw and Spring Boot ​

Supports obtaining real-time service status and configuration information through conversation.

Retrieve Thread Information

Retrieve Configuration Information

Retrieve GC Information

Retrieve Scheduler Information

Role-Level Dynamic Tool Authorization v1.14.3+ ​

TIP

Claw's capability boundaries are determined by role. Administrators naturally have all tools; other roles can only call explicitly authorized Tools — high-risk operations are completely invisible and uncallable for unauthorized roles. This enables Claw to be safely deployed in production environments while providing a personalized AI experience for every user.

In the erupt-ai role management interface, check the available Tools for each role and configure a dedicated system prompt. Changes take effect immediately without a restart. See Role-Level Tool Authorization.

@Tools Reference Table ​

The following lists all built-in @Tool entries in Erupt AI Claw. They can be selectively authorized in the role configuration interface.

System Tools ​

ToolDescription
getCurrentDateTimeGet the current date and time
execShellExecute a shell command (high-risk; recommend granting to admins only)

Spring Boot ​

ToolDescription
getActiveProfilesGet the currently active Spring profiles
getPropertyRead the value of a specified Spring configuration property
getScheduledTasksGet the list of Spring scheduled tasks

JVM Monitoring ​

ToolDescription
getJvmMemoryGet JVM heap memory usage
getJvmUptimeGet JVM uptime
getThreadInfoGet JVM thread details
getGcStatsGet garbage collection statistics

Erupt Data ​

ToolDescription
eruptModelListGet the list of all Erupt models
eruptSchemaGet the field schema of a specified Erupt model
eruptDataQueryQuery the data list of an Erupt model
findEruptDataByPkQuery a single Erupt data record by primary key
insertEruptDataInsert a new record into an Erupt model
updateEruptDataUpdate a record in an Erupt model
deleteEruptDataDelete a record from an Erupt model
getEruptModulesGet the list of loaded Erupt modules
geneEruptCodeGenerate Erupt model code

Erupt User ​

ToolDescription
eruptUserInfoGet the current logged-in user's information
eruptUserPermissionsGet the current user's permission information

Cube Analytics ​

ToolDescription
cubeListList all Cube datasets
cubeMetadataGet the metadata of a specified Cube dataset
cubeQueryExecute a Cube dataset query

Skill Management ​

ToolDescription
listSkillsList all available Skills
getSkillDetailGet details of a specified Skill
readSkillFileRead the content of a Skill file
saveSkillCreate or update a Skill
deleteSkillDelete a specified Skill

Memory Management ​

ToolDescription
listMemoriesList all memory entries
getMemoryGet the content of a specified memory
saveMemorySave a memory entry
deleteMemoryDelete a specified memory

Contributors

The avatar of contributor named as YuePeng YuePeng

Changelog

Released under the Apache-2.0 License.