From governance
Updates agent files to align with the latest template structure and applies any specified changes while preserving unique agent characteristics.
How this skill is triggered — by the user, by Claude, or both
Slash command
/governance:update-agentThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Updates agent files to align with the latest template structure and applies any specified changes while preserving unique agent characteristics.
Updates agent files to align with the latest template structure and applies any specified changes while preserving unique agent characteristics.
This skill systematically updates agent files to maintain consistency with the current template while preserving each agent's unique personality, expertise, and collaboration networks.
What this skill does NOT do:
/create-agent instead)When to REJECT:
/agents directorytemplate:agent is missing or invalidultrathink: you'd perform the following steps
Check the session context for **Agent Teams**: enabled under the "Agent Capabilities" section.
Analyze Requirements
*frontend*)Load Template Reference
Locate Agents
Create Team
update-agent-teamTemplate Validation
Spawn Agent Update Specialists
team_name: "update-agent-team"name: "updater-{N}" (sequential naming)model: "opus"agent_type: "general-purpose"Create and Assign Tasks
Agent Update Task Specification
Each Agent Update Specialist receives:
ultrathink: adopt the Agent Update Specialist mindset
Assignment You're assigned to update agent: [agent name]
Agent Specifications:
Steps
Read Current Agent:
Compare with Template:
Apply Updates:
Clean & Finalize:
Report [IMPORTANT] You MUST return the following execution report (<500 tokens) via SendMessage to team-lead:
status: success|failure|partial
agent: '[agent-name]'
summary: 'Brief description of changes applied'
modifications:
- section: '[section name]'
change: '[what was changed]'
template_compliance: true|false
personality_preserved: true|false
context_level: '[calculated %]' # (input_tokens / context_window_size) from real usage data
issues: ['issue1', 'issue2', ...] # only if problems encountered
<<<
Progress Monitoring
Collect Results
Shutdown Teammates
Delete Team
| Agent Type | Model | Role | Lifecycle |
|---|---|---|---|
| Agent Update Specialist | opus | Updates agent files with template alignment and changes | One per agent file; spawned for Phase 3, retired in Phase 4 |
When Agent Teams are not available, execute the existing workflow:
Analyze Requirements
*frontend*)Load Template Reference
Locate Agents
Template Validation
Delegation
Progress Monitoring
Template Compliance Verification
Change Application Verification
Personality Preservation Check
Output Format:
[✅/❌] Command: update-agent $ARGUMENTS
## Summary
- Execution mode: [team/subagent]
- Agents processed: [count/total]
- Successfully updated: [count]
- Failed updates: [count]
- Template compliance: [PASS/FAIL]
## Actions Taken
1. [Batch processing of agents with results]
2. [Template alignment changes applied]
3. [Custom changes applied (if any)]
## Skills Applied (subagent mode)
- Update Agent Skill: [Status]
## Teammate Results (team mode only)
- Total agents deployed: [count]
- Successful updates: [count]
- Failed updates: [count] (if any)
## Updated Agents
- [agent-name]: [Status] - [Changes applied]
- [agent-name]: [Status] - [Changes applied]
## Issues Found (if any)
- **Agent**: [agent-name]
**Issue**: [Description of problem]
**Resolution**: [Applied fix or manual intervention needed]
/update-agent
# With CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS=1:
# - Creates update-agent-team
# - Spawns parallel Agent Update Specialists (one per agent file)
# - Each specialist updates agent with template alignment
# - Preserves unique characteristics while ensuring template compliance
# - Reports execution mode: team
/update-agent
# Without agent teams:
# - Uses traditional subagent delegation
# - Updates all agents in /agents directory to latest template
# - Reports execution mode: subagent
/update-agent "priya-sharma"
# Updates only the priya-sharma agent file
# Aligns with template while preserving role-specific content
/update-agent "*frontend*"
# Updates all agents with 'frontend' in their filename
# Useful for updating agents with similar roles or expertise
/update-agent --change="Add new security compliance gate"
# Updates all agents and applies additional changes
# Template alignment plus specified modifications
/update-agent "james-mitchell" --change="Update collaboration network to include new DevOps role"
# Updates specific agent with both template alignment and custom changes
# Preserves agent identity while making requested modifications
/update-agent "*-engineer*" --change="Update tool permissions for new security standards"
# Updates all engineer-type agents with security updates
# Efficient for role-based mass updates
/update-agent "non-existent-agent"
# Error: Agent file not found
# Suggestion: Use 'ls agents/' to see available agents
# Alternative: Use '/update-agent' to update all agents
/update-agent --verify
# Validates all agents against current template without making changes
# Reports compliance status and suggests improvements
npx claudepluginhub alvis/.claude --plugin governanceRefines or creates advanced variants of agent definitions by updating skills, tools, capabilities, and version metadata. Use when an agent's skills are outdated, capability gaps exist, or tool requirements change.
Manages lifecycle of agents, skills, rules, and hooks in .claude/: create/update/delete with cross-reference propagation, subagent delegation for edits, and atomic permissions handling.
Edit, update, remove, or list custom agents stored as Markdown files. Shows current config, prompts for changes, and supports multilingual triggers in EN/IT/FR/ES/DE/PT.