From glean-core
Detect repetitive patterns and suggest automation opportunities.
How this agent operates — its isolation, permissions, and tool access model
Agent reference
glean-core:agents/automation-finderThe summary Claude sees when deciding whether to delegate to this agent
Detect repetitive patterns and suggest automation opportunities. - Find repeated commands or actions - Identify workflow patterns - Suggest commands, hooks, or skills | Pattern Type | Example | Automation Type | |-------------|---------|-----------------| | Repeated command | `npm run build` 5x | Command or script | | File boilerplate | Same structure 3x | Scaffold command | | Post-edit validat...
Detect repetitive patterns and suggest automation opportunities.
| Pattern Type | Example | Automation Type |
|---|---|---|
| Repeated command | npm run build 5x | Command or script |
| File boilerplate | Same structure 3x | Scaffold command |
| Post-edit validation | Always lint after edit | Hook |
| Workflow sequence | A → B → C always | Skill |
Analyze this session for automation opportunities.
## Session Actions
{{sessionActions}}
## Your Analysis Tasks
1. **Command Patterns**
- Which commands were run multiple times?
- Which command sequences always appear together?
- Minimum 2 occurrences to suggest
2. **File Creation Patterns**
- Were similar file structures created?
- Is there repeatable boilerplate?
- Could this be a scaffold template?
3. **Workflow Patterns**
- What steps always happen together?
- What validations are always done?
- Could these be automated?
## Output Requirements
Return a JSON array:
```json
[
{
"type": "command|hook|skill",
"name": "suggested-name",
"description": "What it does",
"trigger": "When to run (for hooks: PreToolUse/PostToolUse)",
"implementation": {
"summary": "Brief implementation description",
"template": "Code template or structure"
},
"evidence": {
"occurrences": 5,
"examples": ["example 1", "example 2"]
},
"impact": {
"timeSaved": "~5 min per use",
"errorsPrevented": "Optional: what errors this prevents"
}
}
]
User-triggered actions via /command-name.
---
allowed-tools: Bash, Write, Edit
description: Short description
---
# /command-name
Instructions...
Automatic triggers on tool use.
{
"hooks": {
"PostToolUse": [{
"matcher": "Edit(*.ts)",
"hooks": [{
"type": "command",
"command": "npm run typecheck"
}]
}]
}
}
Complex workflows invokable by name.
---
name: skill-name
description: What it does
---
# Skill instructions
Multi-step workflow...
## Examples
### Session Actions
### Output
```json
[
{
"type": "command",
"name": "new-api",
"description": "Create a new API route with standard loader/action template",
"trigger": "manual",
"implementation": {
"summary": "Creates app/routes/api/{name}/route.ts with loader and action",
"template": "loader + action + types template"
},
"evidence": {
"occurrences": 3,
"examples": ["users", "posts", "comments"]
},
"impact": {
"timeSaved": "~3 min per API route"
}
},
{
"type": "hook",
"name": "auto-typecheck",
"description": "Run typecheck after editing TypeScript files",
"trigger": "PostToolUse:Edit(*.tsx)",
"implementation": {
"summary": "PostToolUse hook that runs npm run typecheck",
"template": "{ \"matcher\": \"Edit(*.tsx)\", \"command\": \"npm run typecheck\" }"
},
"evidence": {
"occurrences": 8,
"examples": ["After UserCard.tsx edit", "After Header.tsx edit"]
},
"impact": {
"timeSaved": "Catches errors immediately",
"errorsPrevented": "Type errors discovered earlier"
}
}
]
npx claudepluginhub lledellebell/glean --plugin glean-notifyScans work sessions for repetitive patterns and recommends automation as skills, commands, or subagents with concrete examples and duplicate checks.
Analyzes Glean data via memory, user activity, and search tools to identify repeated patterns, common queries, and workflows for Claude Code skill automation opportunities.
Transforms kaizen analysis findings into actionable improvements: hook configurations, agent/skill prompt patches, CLAUDE.md updates, and automation scripts.