/attributes:route
Route to specialized agents based on codebase health attributes.
When to Use This Skill
| Use this skill when... | Use another approach when... |
|---|---|
| Have attribute data and want automated remediation | Need to collect attributes first (use /attributes:collect) |
| Want severity-based agent prioritization | Know exactly which agent to use |
| Batch-fixing multiple health findings | Fixing a single specific issue |
Parameters
| Parameter | Description |
|---|---|
--dry-run | Show routing plan without spawning agents |
--focus <category> | Only route for: docs, tests, security, quality, ci |
--min-severity <level> | Minimum severity: critical, high, medium (default), low |
Execution
Step 1: Load Attributes
Read attribute data from one of:
.claude/attributes.json(if exists)- Run
/attributes:collectto generate fresh data
Step 2: Filter and Prioritize
- Filter attributes by
--focuscategory if specified - Filter by
--min-severitythreshold - Group by target agent from the
actionsarray - Calculate priority per agent using weights: critical=4, high=3, medium=2, low=1
- Sort agents by descending priority
Step 3: Display Routing Plan
Show the routing plan:
Attribute-Based Routing
Based on attribute analysis, prioritize these agents in order:
1. security (priority 7): .env file committed; No security scanning in CI
2. test_runner (priority 3): No test directory or test files found
3. docs (priority 2): Missing CLAUDE.md
If --dry-run, stop here.
Step 4: Execute Routing
For each agent in priority order, spawn the agent with its findings as context:
| Attribute Category | Agent | Condition |
|---|---|---|
| security (critical/high) | security-audit | Always route first |
| tests | test | When test gaps found |
| quality (high+) | refactor | Anti-patterns exceed threshold |
| quality (medium) | review | Code review suggestions |
| docs | docs | Documentation gaps |
| ci | configure | CI configuration gaps |
When spawning each agent, include:
- The specific findings for that agent
- Whether each finding is auto-fixable
- The severity level
Step 5: Report
After all agents complete, summarize:
- Which agents were invoked
- Which findings were addressed
- Remaining findings that need manual attention
Agentic Optimizations
| Context | Command |
|---|---|
| Preview routing | /attributes:route --dry-run |
| Security only | /attributes:route --focus security |
| Critical only | /attributes:route --min-severity critical |
| Full remediation | /attributes:route |