Operations

Operations are atomic coding tasks with configurable executor backends.

Structure

Field	Description
id	Unique identifier
name	Human-readable name
description	What this operation does
executor	Executor configuration
tags	Categorization labels

Executor Configuration

Agent Executor

Uses an AI agent (Claude or Codex) to execute the operation.

{
  "executor": {
    "type": "agent",
    "agentMode": "skill",
    "agent": "local-claude",
    "skillId": "my-skill",
    "allowedTools": ["Read", "Write", "Edit"]
  }
}

Field	Description
type	"agent"
agentMode	"skill" (uses a skill) or "prompt" (direct prompt)
agent	"local-claude" or "codex"
skillId	Skill ID to use (when agentMode: "skill")
prompt	Direct prompt text (when agentMode: "prompt")
allowedTools	Tools the agent can use

Script Executor

Runs a custom script.

{
  "executor": {
    "type": "script",
    "command": "eslint --format json ."
  }
}

Creating an Operation

Via the REST API

curl -X PUT http://localhost:9433/api/operations/lint-check \
  -H 'Content-Type: application/json' \
  -d '{
    "name": "Lint Check",
    "description": "Run linting on the input files",
    "executor": {
      "type": "agent",
      "agentMode": "prompt",
      "agent": "local-claude",
      "prompt": "Analyze the provided code for style issues."
    },
    "tags": ["lint", "quality"]
  }'

Agent Tools

When using agent executors, you can restrict which tools the agent has access to:

Tool	Description
Read	Read file contents
Write	Write/create files
Edit	Edit existing files
Glob	List files by pattern
Grep	Search file contents
Bash	Execute shell commands
MultiEdit	Edit multiple files