Smart City Ewaste Recycling

SKILL.md — Smart City E-Waste Recycling Skill Registry

Overview

This document provides the complete skill registry for the smart-city-ewaste-recycling harness, including registration protocols, resolution mechanisms, execution flows, validation schemas, and input/output specifications.

Skill Registration

Registration Protocol

The skill registers itself through the following mechanisms:

1. Primary Registration: SKILL.md (this file) - serves as the canonical registry
2. Frontmatter Registry: YAML block at top of skills/main.md
3. Manifest Entry: Library-level progression tracking in D:\972026\progression.json
4. Hook Integration: Event-driven hooks in /hooks directory

Registration Metadata

skill_identifier: smart-city-ewaste-recycling
version: 2.0.0
last_updated: 2026-07-27
status: production-ready
phase: Phase 5 (Integration & Polish)
library_standard: 1.0
base_directory: D:\972026\242-smart-city-ewaste-recycling

Skill Resolution

Resolution Mechanism

When a user invokes /smart-city-ewaste-recycling or provides relevant input, the skill resolves through:

1. Trigger Detection: Description matching in SKILL.md frontmatter
2. Context Loading: Multi-level progressive disclosure system
3. Sub-Skill Resolution: Recursive resolution of dependent sub-skills
4. Resource Binding: Dynamic binding of bundled resources

Resolution Priority

1. User explicit invocation (/smart-city-ewaste-recycling)
2. Semantic trigger detection (description matching)
3. Domain keyword detection (e-waste, urban recycling, WEEE, etc.)
4. Context inference (smart city, circular economy, electronics sustainability)

Context Loading Levels

Level 1 - Metadata (~100 words)

• Skill name, description, version
• Always loaded in Claude context

**Level 2 - SKILL.md Body ( B[Pre-Flight: Language Detection] B --> C[Step 1: sub-gather-requirements] C --> D{Gate G0: Object Confirmed?} D -->|No| C D -->|Yes| E[Step 2: sub-evidence-collector] E --> F{Gate G1: Data Retrieved?} F -->|No (2x)| G[Degradation Level 1] F -->|Yes| H[Step 3: sub-core-analysis] H --> I{Gate G2: Analysis Complete?} I -->|No (2x)| J[Degradation Level 2] I -->|Yes| K[Step 4: sub-knowledge-updater] K --> L{Gate G3: Evidence Found?} L -->|No (2x)| M[Degradation Level 3] L -->|Yes| N[Step 5: sub-advisor] N --> O{Gate G4: Conclusion Valid?} O -->|No (2x)| P[Degradation Level 4] O -->|Yes| Q[Quality Gate Review] Q --> R{All Gates Pass?} R -->|Yes| S[Deliver Output] R -->|No| T[Auto-Fix Loop] T --> Q

### Sub-Skill Execution Matrix

| Sub-Skill | Tool Access | Output Format | Retry Logic |
|-----------|-------------|----------------|-------------|
| sub-gather-requirements | Conversation | Structured requirements | Re-prompt user |
| sub-evidence-collector | WebSearch, WebFetch, Read | Evidence bundle | Fallback to KB |
| sub-core-analysis | Read, WebFetch, Reasoning | Scorecard/analysis | Fallback to frameworks |
| sub-knowledge-updater | Read, WebSearch | Citations + tier labels | Flag gaps |
| sub-advisor | Skill, Reasoning | Conclusion + disclosure | Force to template |

## Skill Validation

### Validation Schemas

**Input Validation:**

```json
{
  "user_query": {
    "type": "string",
    "min_length": 10,
    "required": true,
    "validation": "non_empty_meaningful_content"
  },
  "language": {
    "type": "string",
    "enum": ["en", "vi", "other"],
    "default": "en"
  }
}

Output Validation:

{
  "report_structure": {
    "required_sections": [
      "Executive Summary",
      "Inputs & Scope",
      "Evidence Collected",
      "Analysis / Scorecard",
      "Control / Action Plan",
      "Academic Evidence",
      "Verdict / Conclusion",
      "Key Risks",
      "Evidence Chain",
      "Recommended Actions",
      "Disclosure / Limitations"
    ],
    "validation": "all_sections_present_and_populated"
  },
  "quality_gates": {
    "universal_gates": ["U1", "U2", "U3", "U4", "U5", "U6"],
    "domain_gates": ["G1", "G2", "G3", "G4"],
    "pass_criteria": "all_gates_must_pass"
  }
}

Quality Gate Enforcement

# Gate validation schema
gates = {
    "U1": {"check": "source_count >= 3", "academic": ">= 1", "fix": "append_missing"},
    "U2": {"check": "disclosure_position", "before": "recommendation", "fix": "prepend_disclosure"},
    "U3": {"check": "tier_labels_present", "tiers": [1,2,3,4], "fix": "annotate_tiers"},
    "U4": {"check": "language_match", "fix": "translate_output"},
    "U5": {"check": "template_compliance", "fix": "reformat_to_template"},
    "U6": {"check": "claim_traceability", "fix": "flag_unsupported"},
    
    "G1": {"check": "flows_mapped", "fix": "map_e_waste_flows"},
    "G2": {"check": "collection_recycling_specified", "fix": "specify_collection_recycling"},
    "G3": {"check": "hazards_mitigated", "fix": "mitigate_hazards"},
    "G4": {"check": "policy_compliance", "fix": "ensure_compliance"}
}

Sub-Skill Registry

Sub-Skill Resolution Table

| Sub-Skill File | Name | Purpose | Dependencies | Tools Required | |----------------|------|---------|--------------|----------------| | sub-gather-requirements.md | Requirements Intake | Clarify analysis parameters | None | Conversation | | sub-evidence-collector.md | Evidence Collection | Fetch authoritative data | Requirements output | WebSearch, WebFetch, Read | | sub-core-analysis.md | Core Analysis | Domain-specific analysis | Evidence bundle | Read, WebFetch, Reasoning | | sub-knowledge-updater.md | Knowledge Query | Surface academic evidence | Topic keywords | Read, WebSearch | | sub-advisor.md | Synthesis & Recommendation | Final conclusion | All prior outputs | Skill, Reasoning |

Sub-Skill JSON Schemas

Requirements Output Schema:

{
  "requirements": {
    "object": "string",
    "scope": "string",
    "timeframe": "string",
    "available_inputs": ["string"],
    "target_audience": "string",
    "language": "en|vi",
    "analysis_type": "string"
  }
}

Evidence Bundle Schema:

{
  "evidence_bundle": {
    "current_data": {
      "source": "string",
      "date": "ISO8601",
      "content": "object"
    },
    "authoritative_docs": [{
      "title": "string",
      "url": "string",
      "tier": 1|2|3|4,
      "key_points": ["string"]
    }],
    "recent_news": [{
      "title": "string",
      "url": "string",
      "date": "ISO8601",
      "summary": "string"
    }],
    "reference_benchmarks": {
      "metric": "value",
      "source": "string"
    }
  }
}

Error Handling & Degradation

Error Types and Recovery

| Error Type | Recovery Strategy | Degradation Level | |------------|-------------------|-------------------| | Network Failure | Fallback to knowledge base | Level 1-2 | | Source Unavailable | Alternative sources + flag | Level 1 | | Timeout | Retry with reduced scope | Level 2 |

• Parse Error | Skip problematic content | Level 2-3 |

| Language Mismatch | Auto-translate | Level 1 | | Quality Gate Failure | Auto-fix or explicit flag | Level 3-4 | | Sub-Skill Failure | Graceful degradation | Level 3 | | Complete Failure | Limitation disclosure | Level 4 |

Degradation Level Indicators

**LIMITATION LEVEL 1:** Some real-time data unavailable; using knowledge base.
**LIMITATION LEVEL 2:** Network issues; using cached data with reduced scope.
**LIMITATION LEVEL 3:** Critical sources unavailable; analysis based on partial data.
**LIMITATION LEVEL 4:** Severe limitations; results inconclusive. Manual review required.

Hooks System

Hook Registration

Hooks are registered in /hooks/ directory with the following naming convention:

hooks/
├── pre-execution/       # Before skill execution
├── post-execution/      # After skill execution
├── on-error/           # Error handling hooks
└── lifecycle/          # Lifecycle management hooks

Hook Execution Order

1. pre-execution/language-detection.md
2. pre-execution/context-validation.md
3. [Skill execution - Steps 1-6]
4. post-execution/output-validation.md
5. post-execution/quality-gate-check.md
6. lifecycle/knowledge-update.md (if triggered)

Hook Schema

hook_name: description
trigger: event_type
priority: number
condition: boolean_expression
action: function_call
output: schema_definition

Performance Optimization

Context Window Management

• Progressive Disclosure: Load only necessary content
• Token Budgeting: Allocate 2K tokens per sub-skill maximum
• Caching Strategy: Cache frequently accessed knowledge base sections
• Cleanup Protocol: Clear resolved sub-skills after execution

Token Consumption Tracking

# Token tracking schema
token_usage = {
    "metadata": "~100",
    "skill_body": "~500 lines",
    "sub_skills": {
        "sub-gather-requirements": "2K max",
        "sub-evidence-collector": "4K max",
        "sub-core-analysis": "8K max",
        "sub-knowledge-updater": "2K max",
        "sub-advisor": "4K max"
    },
    "quality_gates": "1K max",
    "total_estimated": "21K max per session"
}

Integration Points

External Tool Integration

WebSearch: Live domain news, reports, standards updates WebFetch: Scrape authoritative sources Read: Read knowledge base and reference files Write: Append to knowledge base (knowledge pipeline) Skill: Invoke sub-skills recursively Bash: Execute knowledge pipeline (scheduled)

Knowledge Pipeline Integration

# Weekly academic update (Mondays 8:00 AM)
0 8 * * 1 python tools/knowledge_updater.py >> logs/knowledge_update.log 2>&1

# Daily news update (Daily 7:00 AM)
0 7 * * * python tools/knowledge_updater.py --news-only >> logs/knowledge_news.log 2>&1

Version History

| Version | Date | Changes | |---------|------|---------| | 2.0.0 | 2026-07-27 | Production-grade architecture with flexible agent system | | 1.0.0 | 2026-07-13 | Initial production release |

Maintenance Guidelines

Update Protocol

1. Schema Updates: Update JSON schemas in this file first
2. Sub-Skill Changes: Update sub-skill resolution table
3. New Hooks: Register in hooks system section
4. Version Bump: Update version history and all frontmatter

Testing Protocol

# Run full test suite
python tools/run_test_scenarios.py --all

# Test knowledge pipeline
python tools/test_knowledge_updater.py

# Validate project structure
python tools/validate_project.py

Support & Debugging

Debug Mode

# Enable debug logging
export SKILL_DEBUG=1
/smart-city-ewaste-recycling --debug [query]

Common Issues

| Issue | Cause | Solution | |-------|-------|----------| | Skill not triggering | Description mismatch | Update SKILL.md frontmatter | | Quality gate failures | Insufficient evidence | Run knowledge updater | | Language detection errors | Unusual input | Manual language override | | Sub-skill timeouts | Network issues | Check degradation levels |

---

End of SKILL.md Registry Document

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: dungnotnull
• Source: dungnotnull/smart-city-ewaste-recycling-agent-skill
• License: MIT

Install and usage instructions live in the source repository linked above.