Specialized subagents for CCPM operations and development workflows.
/plugin marketplace add duongdev/ccpm/plugin install ccpm@duongdev-ccpm-marketplaceSpecialized subagents for CCPM operations and development workflows.
CCPM agents are specialized components that handle specific domains of functionality. They are invoked automatically by commands, hooks, or explicitly via the Task tool, providing focused expertise and optimized token usage through caching and batching.
These agents handle project management integrations and provide centralized API operations with caching.
| Agent | Purpose |
|---|---|
| linear-operations | Central handler for all Linear API operations with session-level caching |
| pm-operations-orchestrator | Lightweight coordinator for multi-PM system operations with lazy loading |
| jira-operations | Jira API operations with caching and Markdown/ADF conversion |
| confluence-operations | Confluence API operations with content-aware caching |
These agents manage project detection, configuration, and context across CCPM workflows.
| Agent | Purpose |
|---|---|
| project-detector | Detects active project and subproject from git remote, working directory, or patterns |
| project-config-loader | Loads and validates project configuration from CCPM config file |
| project-context-manager | Manages active project context including setting, displaying, and switching projects |
These agents handle implementation tasks across different technology domains.
| Agent | Purpose |
|---|---|
| frontend-developer | React/UI implementation with design system integration and Tailwind styling |
| backend-architect | API design, NestJS implementation, database operations, and authentication |
| tdd-orchestrator | Test-driven development workflow orchestration with coverage requirements |
| code-reviewer | Automated code review covering security, bugs, performance, and quality |
| code-quality-enforcer | Automated quality validation running lint/type/test checks on changed files |
| debugger | Systematic debugging and issue investigation with root cause analysis |
| security-auditor | Security vulnerability assessment covering OWASP Top 10 and compliance |
| Agent | Purpose |
|---|---|
| pm:ui-designer | UI/UX design with wireframes, design system analysis, and developer specifications |
These agents help users work with Claude Code and extend CCPM functionality.
| Agent | Purpose |
|---|---|
| claude-code-guide | Answers questions about Claude Code features, settings, hooks, skills, and MCP |
| ccpm-developer | Creates and extends CCPM commands, agents, skills, and hooks |
| ccpm-troubleshooter | Diagnoses and resolves Claude Code and CCPM configuration issues |
The smart-agent-selector hook automatically scores and suggests agents based on:
Agents scoring above threshold are automatically invoked or suggested.
Invoke agents explicitly using the Task tool:
Task({
subagent_type: 'ccpm:frontend-developer',
prompt: `
## Task
Create login form component
## Context
- Issue: PSN-123
- Branch: feature/psn-123-auth
## Requirements
- TypeScript strict mode
- Tailwind styling
- Accessible (a11y)
`
});
Commands delegate to agents internally based on task type:
| Task Type | Agent |
|---|---|
| UI/Frontend | frontend-developer |
| API/Backend | backend-architect |
| Testing | tdd-orchestrator |
| Code Review | code-reviewer |
| Quality Checks | code-quality-enforcer |
| Debugging | debugger |
| Security | security-auditor |
| Design | pm:ui-designer |
The linear-operations agent is the central handler for all Linear API interactions. It provides significant token reduction through caching and batching.
The Linear MCP uses specific parameter names that differ between operations:
| Tool | Parameter | Example |
|---|---|---|
get_issue | id | { id: "WORK-26" } |
update_issue | id | { id: "WORK-26", state: "..." } |
create_comment | issueId | { issueId: "WORK-26", body: "..." } |
list_comments | issueId | { issueId: "WORK-26" } |
create_issue | team, title | { team: "Engineering", title: "..." } |
For non-blocking updates, use background execution:
# Post comment (fire-and-forget)
./scripts/linear-background-ops.sh comment PSN-123 "Progress update"
# Update status (fire-and-forget)
./scripts/linear-background-ops.sh update-status PSN-123 "In Progress"
Use blocking calls only when you need the result:
get_issue - Need the data to continuecreate_issue - Need the issue IDupdate_checklist_items - Need progress for displayThe pm-operations-orchestrator provides multi-PM system coordination with:
| Operation | Purpose |
|---|---|
lazy_gather_context | Gather context from multiple PM systems |
smart_delegate | Route single operation to appropriate subagent |
batch_parallel_execute | Execute multiple operations with dependencies |
cache_status | Report unified cache metrics |
# Agent Name
**Specialized agent for [specific domain]**
## Purpose
What this agent does and why.
## Capabilities
- Capability 1
- Capability 2
## Input Contract
```yaml
task:
type: string
description: string
context:
issueId: string?
result:
status: "success" | "partial" | "blocked"
filesModified: string[]
summary: string
Code examples and patterns.
How commands invoke this agent.
Concrete usage examples.
### Registration
1. Create agent file in `agents/` directory
2. Add to `.claude-plugin/plugin.json` agents array:
```json
{
"agents": [
"./agents/existing-agent.md",
"./agents/new-agent.md"
]
}
Commands
│
├─> linear-operations (Linear API)
│
├─> pm-operations-orchestrator
│ ├─> linear-operations
│ ├─> jira-operations
│ └─> confluence-operations
│
├─> Development Agents
│ ├─> frontend-developer
│ ├─> backend-architect
│ ├─> tdd-orchestrator
│ └─> ...
│
└─> Project Agents
├─> project-detector
├─> project-config-loader
└─> project-context-manager
| Category | Agent Count | Purpose |
|---|---|---|
| Linear/PM Operations | 4 | API operations with caching |
| Project Management | 3 | Project detection and context |
| Development | 7 | Implementation across domains |
| Design | 1 | UI/UX design assistance |
| CCPM Support | 3 | Claude Code and CCPM help |
| Total | 18 |
You are an elite AI agent architect specializing in crafting high-performance agent configurations. Your expertise lies in translating user requirements into precisely-tuned agent specifications that maximize effectiveness and reliability.