From utils
Routes natural language requests to skills and workflows by parsing intent and matching patterns for SDLC phases, gate checks, artifacts, security reviews, and marketing tasks.
npx claudepluginhub jmagly/aiwg --plugin utilsThis skill uses the workspace's default tool permissions.
Route natural language requests to appropriate skills and workflows.
Searches, retrieves, and installs Agent Skills from prompts.chat registry using MCP tools like search_skills and get_skill. Activates for finding skills, browsing catalogs, or extending Claude.
Searches prompts.chat for AI prompt templates by keyword or category, retrieves by ID with variable handling, and improves prompts via AI. Use for discovering or enhancing prompts.
Guides Next.js Cache Components and Partial Prerendering (PPR) with cacheComponents enabled. Implements 'use cache', cacheLife(), cacheTag(), revalidateTag(), static/dynamic optimization, and cache debugging.
Route natural language requests to appropriate skills and workflows.
This skill is the meta-router - it processes natural language and routes to other skills. It should be consulted when no explicit skill trigger is matched.
This skill interprets natural language requests and routes them to the appropriate skills or workflows by:
When triggered, this skill:
Parses user intent:
Matches to skills:
Routes request:
Handles follow-ups:
sdlc_patterns:
phase_transitions:
patterns:
- "transition to {phase}"
- "move to {phase}"
- "start {phase}"
- "begin {phase}"
- "ready for {phase}"
routes_to: flow-{from}-to-{to}
extract: [from_phase, to_phase]
gate_checks:
patterns:
- "check gate"
- "can we transition"
- "ready for {phase}"
- "validate {gate}"
- "gate check"
routes_to: gate-evaluation
extract: [gate_name, phase]
project_status:
patterns:
- "where are we"
- "what's next"
- "project status"
- "current phase"
- "status update"
routes_to: project-awareness
action: status_report
risk_management:
patterns:
- "update risks"
- "risk review"
- "new risk"
- "mitigate {risk}"
routes_to: risk-cycle
extract: [risk_id, action]
security_review:
patterns:
- "security review"
- "run security"
- "threat model"
- "security scan"
routes_to: security-assessment
extract: [scope]
artifact_patterns:
create:
patterns:
- "create {artifact}"
- "generate {artifact}"
- "new {artifact}"
- "draft {artifact}"
routes_to: artifact-orchestration
extract: [artifact_type]
review:
patterns:
- "review {artifact}"
- "check {artifact}"
- "validate {artifact}"
routes_to: artifact_type_specific_review
extract: [artifact_type]
trace:
patterns:
- "trace {requirement}"
- "what implements {id}"
- "coverage for {id}"
routes_to: traceability-check
extract: [requirement_id]
marketing_patterns:
brand_review:
patterns:
- "brand review"
- "check brand compliance"
- "is this on-brand"
- "brand audit"
routes_to: brand-compliance
extract: [asset_path]
approval:
patterns:
- "submit for approval"
- "approval workflow"
- "get sign-off"
- "approval status"
routes_to: approval-workflow
extract: [asset_id]
performance:
patterns:
- "how are we doing"
- "marketing report"
- "performance summary"
- "campaign results"
routes_to: performance-digest
extract: [period, channel]
competitive:
patterns:
- "competitor analysis"
- "what are competitors doing"
- "competitive landscape"
routes_to: competitive-intel
extract: [competitor_name]
research_patterns:
lookup:
patterns:
- "look up {topic}"
- "what does {thing} do"
- "how does {thing} work"
- "find documentation for {topic}"
- "search for {query}"
- "check the docs for {topic}"
- "read about {topic}"
- "research {topic}"
routes_to: research-investigation
extract: [topic, scope]
rule_activation: research-before-decision
codebase_exploration:
patterns:
- "how is {feature} implemented"
- "where is {thing} defined"
- "find {pattern} in the codebase"
- "what pattern does this project use for {thing}"
- "show me examples of {pattern}"
routes_to: codebase-exploration
extract: [feature, pattern]
rule_activation: research-before-decision
error_investigation:
patterns:
- "why is {thing} failing"
- "investigate {error}"
- "debug {issue}"
- "what's causing {problem}"
- "root cause of {error}"
routes_to: error-diagnosis
extract: [error, context]
rule_activation: research-before-decision
planning_patterns:
approach_planning:
patterns:
- "plan how to {task}"
- "design approach for {task}"
- "think through {task}"
- "strategize {task}"
- "how should I approach {task}"
- "what's the best way to {task}"
routes_to: approach-planning
extract: [task, constraints]
implementation_planning:
patterns:
- "plan the implementation of {feature}"
- "break down {task}"
- "outline the steps for {task}"
- "what do I need to do for {task}"
routes_to: implementation-planning
extract: [feature, scope]
decision_support:
patterns:
- "help me decide between {options}"
- "compare {a} and {b}"
- "trade-off analysis for {topic}"
- "which should I choose"
- "pros and cons of {approach}"
routes_to: decision-support
extract: [decision_topic, options]
clarification_patterns:
re_read:
patterns:
- "re-read my instructions"
- "that's not what I asked"
- "I said {correction}"
- "go back and read what I wrote"
- "you missed {thing}"
- "I already told you {thing}"
routes_to: instruction-reparse
rule_activation: instruction-comprehension
priority: high
confusion:
patterns:
- "I'm confused about {topic}"
- "can you explain {thing}"
- "what do you mean by {thing}"
- "I don't understand {thing}"
routes_to: clarification
extract: [topic]
correction:
patterns:
- "no, I meant {correction}"
- "not {wrong}, use {right}"
- "change that to {correction}"
- "that's wrong, it should be {correction}"
routes_to: instruction-correction
rule_activation: instruction-comprehension
priority: high
extract: [correction]
utility_patterns:
decision:
patterns:
- "help me decide"
- "compare options"
- "trade-off analysis"
- "which should I choose"
routes_to: decision-support
extract: [decision_topic]
test_coverage:
patterns:
- "test coverage"
- "what's not tested"
- "coverage report"
routes_to: test-coverage
extract: [scope]
incident:
patterns:
- "production issue"
- "system down"
- "incident"
- "P0"
- "SEV1"
routes_to: incident-triage
priority: urgent
config:
patterns:
- "validate config"
- "check setup"
- "config issues"
routes_to: config-validator
action_verbs:
create:
words: [create, generate, new, draft, make, build, write]
intent: creation
review:
words: [review, check, validate, verify, audit, assess]
intent: validation
update:
words: [update, change, modify, edit, revise]
intent: modification
delete:
words: [delete, remove, deprecate, archive]
intent: removal
analyze:
words: [analyze, understand, explain, investigate, explore]
intent: analysis
research:
words: [research, look up, find out, search for, read about, check docs]
intent: investigation
rule_activation: research-before-decision
plan:
words: [plan, design, strategize, outline, think through, approach]
intent: planning
clarify:
words: [clarify, re-read, explain, what did I, I said, not what I asked]
intent: clarification
rule_activation: instruction-comprehension
transition:
words: [transition, move, progress, advance, start, begin]
intent: workflow_progression
compare:
words: [compare, contrast, versus, vs, trade-off]
intent: comparison
object_mappings:
artifacts:
sad: software-architecture-document
adr: architecture-decision-record
test_plan: test-plan
requirements: software-requirements-spec
threat_model: threat-model
phases:
inception: inception
elaboration: elaboration
construction: construction
transition: transition
production: production
gates:
lom: lifecycle-objective-milestone
abm: architecture-baseline-milestone
ioc: initial-operational-capability
prm: product-release-milestone
confidence_levels:
high:
threshold: 0.85
action: route_directly
example: "create SAD" → artifact-orchestration (SAD)
medium:
threshold: 0.60
action: confirm_before_routing
example: "review this" → "Did you mean brand review or code review?"
low:
threshold: 0.40
action: offer_options
example: "help" → "Here are things I can help with..."
none:
threshold: 0.0
action: clarify
example: "do the thing" → "I'm not sure what you'd like to do. Can you clarify?"
routing_response:
confidence: high
matched_skill: artifact-orchestration
extracted_params:
artifact_type: sad
message: "Creating Software Architecture Document..."
next_action: invoke_skill
routing_response:
confidence: medium
candidates:
- skill: brand-compliance
confidence: 0.72
reason: "brand" keyword matched
- skill: code-reviewer
confidence: 0.65
reason: "review" action detected
message: "I found multiple possible actions. Did you mean..."
next_action: await_clarification
routing_response:
confidence: low
suggestions:
- category: Documentation
actions: [create SAD, generate ADR, draft requirements]
- category: Review
actions: [brand review, code review, security review]
- category: Status
actions: [project status, coverage report, risk review]
message: "I'm not sure what you'd like to do. Here are some options..."
next_action: await_selection
User: "Generate the architecture document"
Skill routes:
- Intent: create
- Object: architecture document → SAD
- Confidence: 0.95 (high)
Output:
"Routing to artifact-orchestration to generate Software Architecture Document.
Starting SAD generation..."
[Invokes artifact-orchestration with artifact_type=sad]
User: "Review the landing page"
Skill analyzes:
- Intent: review
- Object: landing page
- Possible skills: brand-compliance, qa-protocol, code-reviewer
- Confidence: 0.65 (medium)
Output:
"I found a few review options for 'landing page':
1. **Brand Review** - Check brand compliance
2. **QA Review** - Check quality and specs
3. **Content Review** - Check copy and messaging
Which would you like?"
User: "Help with the project"
Skill analyzes:
- Intent: unclear
- Object: project (generic)
- Confidence: 0.30 (low)
Output:
"I can help with many things! Here are some common actions:
**Project Status**
- Where are we? → Project status
- What's next? → Phase recommendations
**Documentation**
- Create SAD → Architecture document
- Generate ADR → Decision record
**Reviews**
- Security review → Threat assessment
- Brand review → Compliance check
What would you like to do?"
User: (After discussing a bug) "Create a ticket for this"
Skill analyzes:
- Context: Bug discussion
- Intent: create
- Object: ticket
- Inference: incident or defect ticket
- Confidence: 0.80 (high with context)
Output:
"Based on our discussion, I'll create an incident ticket for the authentication issue.
Routing to incident-triage..."
This skill is the central router and integrates with all other skills:
Add custom routing patterns in .aiwg/config/routing.yaml:
custom_patterns:
- trigger: "standup update"
routes_to: project-awareness
action: standup_summary
- trigger: "weekly report"
routes_to: performance-digest
params:
period: weekly
priority_overrides:
# Urgent patterns always match first
urgent:
- "production down"
- "SEV1"
- "security breach"
# Exact matches before fuzzy
exact_first: true
# Prefer skill in current context
context_boost: 0.15
.aiwg/logs/routing/.aiwg/analytics/routing-patterns.json