Lead Free Heat Resistant Ceramics

Lead-Free Heat-Resistant Ceramic Glaze & Body Formulation Skill

Overview

This skill is a production-grade harness for Ceramic Materials & Lead-Free Glaze Engineering analysis. It transforms Claude into a domain expert that delivers structured, evidence-backed outputs by combining real-time data aggregation, recognized domain methods, and academic research into a single orchestrated workflow.

Skill Registration & Resolution

Registration Schema

Skills are registered in the skills/ directory with the following schema:

---
name: 
description: 
version: 
author: 
license: 
compatibility: 
---

Skill Resolution Process

1. Input Parsing: Analyze user input for domain keywords and intent
2. Skill Matching: Compare against registered skill descriptions
3. Priority Scoring: Rank matches by relevance score
4. Context Loading: Load the matched skill's SKILL.md body
5. Sub-Skill Resolution: Resolve required sub-skills recursively

Resolution Fallback Chain

Primary Skill (main.md) → Sub-Skills (sub-*.md) → References → Tools

If a sub-skill fails to resolve:

• Log the failure with context
• Attempt degradation to available sub-skills
• Flag limitation in output
• Continue with reduced capability

Skill Execution Model

Execution Flow

[User Input] → [Language Detection] → [Requirement Gathering] → [Evidence Collection] → [Core Analysis] → [Knowledge Integration] → [Synthesis] → [Quality Gate] → [Output]

Execution States

| State | Description | Transitions | |-------|-------------|-------------| | INITIALIZED | Skill loaded and ready | → RUNNING | | RUNNING | Actively processing | → COMPLETE, ERROR, DEGRADED | | DEGRADED | Running with limitations | → COMPLETE, ERROR | | COMPLETE | Finished successfully | → INITIALIZED | | ERROR | Failed with error | → INITIALIZED |

State Machine

          ┌─────────────┐
          │ INITIALIZED │◀─────────┐
          └──────┬──────┘          │
                 │                 │
                 ▼                 │
          ┌─────────────┐          │
          │  RUNNING    │          │
          └──────┬──────┘          │
                 │                 │
        ┌────────┴────────┐        │
        ▼                 ▼        │
   ┌─────────┐      ┌──────────┐ │
   │ DEGRADED│      │  ERROR   │ │
   └────┬────┘      └──────────┘ │
        │                          │
        └──────────────────────────┘
                 │
                 ▼
          ┌─────────────┐
          │  COMPLETE   │
          └─────────────┘

Sub-Skill Registry

Available Sub-Skills

| Skill Name | Purpose | Input Schema | Output Schema | |------------|---------|--------------|---------------| | sub-gather-requirements | Intake & scope clarification | {raw_message, provided_materials} | {object, scope, timeframe, inputs, audience, language, analysis_type} | | sub-evidence-collector | Real-time data aggregation | {requirements} | {current_data, authoritative_docs, recent_news, reference_benchmarks} | | sub-core-analysis | Domain-specific analysis | {ware, use, firing, materials, language} | {body_formulation, glaze_formulation, thermal_expansion, leaching_test, firing_schedule} | | sub-knowledge-updater | Academic evidence retrieval | {topic_keywords} | {citations: [{title, authors, year, tier, source}]} | | sub-advisor | Synthesis & recommendation | {analysis, evidence, knowledge} | {verdict, scenarios, risks, evidence_chain, remediation, disclosure} |

Sub-Skill Dependency Graph

sub-gather-requirements (no dependencies)
        │
        ▼
sub-evidence-collector (depends: sub-gather-requirements)
        │
        ▼
sub-core-analysis (depends: sub-gather-requirements, sub-evidence-collector)
        │
        ▼
sub-knowledge-updater (depends: sub-core-analysis)
        │
        ▼
sub-advisor (depends: sub-core-analysis, sub-evidence-collector, sub-knowledge-updater)
        │
        ▼
[Quality Gate] (depends: all previous)

Tool System

Tool Definitions

WebSearch Tool

{
  "name": "WebSearch",
  "description": "Search live domain news and reports",
  "parameters": {
    "query": {"type": "string", "required": true},
    "max_results": {"type": "integer", "default": 10},
    "recency": {"type": "string", "enum": ["oneDay", "oneWeek", "oneMonth", "noLimit"]}
  },
  "timeout": 30,
  "retry_policy": {"max_attempts": 3, "backoff": "exponential"}
}

WebFetch Tool

{
  "name": "WebFetch",
  "description": "Retrieve content from authoritative sources",
  "parameters": {
    "url": {"type": "string", "required": true},
    "format": {"type": "string", "enum": ["markdown", "text"], "default": "markdown"}
  },
  "timeout": 60,
  "retry_policy": {"max_attempts": 2, "backoff": "fixed"}
}

Read Tool

{
  "name": "Read",
  "description": "Read files from filesystem",
  "parameters": {
    "file_path": {"type": "string", "required": true},
    "offset": {"type": "integer", "default": 0},
    "limit": {"type": "integer", "default": 2000}
  }
}

Write Tool

{
  "name": "Write",
  "description": "Write files to filesystem",
  "parameters": {
    "file_path": {"type": "string", "required": true},
    "content": {"type": "string", "required": true}
  }
}

Skill Tool

{
  "name": "Skill",
  "description": "Invoke sub-skills",
  "parameters": {
    "skill": {"type": "string", "required": true},
    "args": {"type": "string", "required": false}
  }
}

Tool Execution Handlers

Each tool has an execution handler that:

1. Validates input parameters against schema
2. Applies rate limiting and timeout controls
3. Executes the tool function
4. Captures output and telemetry
5. Handles errors with graceful degradation
6. Returns structured response

Input/Output JSON Schemas

Input Schema

{
  "$schema": "http://json-schema.org/draft-07/schema#",
  "type": "object",
  "properties": {
    "raw_message": {
      "type": "string",
      "description": "User's raw input message"
    },
    "language": {
      "type": "string",
      "enum": ["en", "vi", "auto"],
      "default": "auto"
    },
    "context": {
      "type": "object",
      "properties": {
        "provided_materials": {"type": "array"},
        "constraints": {"type": "object"},
        "timeframe": {"type": "string"},
        "target_audience": {"type": "string"}
      }
    }
  },
  "required": ["raw_message"]
}

Output Schema

{
  "$schema": "http://json-schema.org/draft-07/schema#",
  "type": "object",
  "properties": {
    "language": {
      "type": "string",
      "enum": ["en", "vi"]
    },
    "report_sections": {
      "type": "object",
      "properties": {
        "executive_summary": {"type": "string"},
        "inputs_scope": {"type": "object"},
        "evidence_collected": {"type": "array"},
        "analysis_scorecard": {"type": "object"},
        "action_control_plan": {"type": "object"},
        "academic_evidence": {"type": "array"},
        "disclosure": {"type": "string"},
        "recommendation": {"type": "object"}
      },
      "required": ["executive_summary", "disclosure", "recommendation"]
    },
    "quality_gates": {
      "type": "object",
      "properties": {
        "U1": {"type": "boolean"},
        "U2": {"type": "boolean"},
        "U3": {"type": "boolean"},
        "U4": {"type": "boolean"},
        "U5": {"type": "boolean"},
        "U6": {"type": "boolean"},
        "G1": {"type": "boolean"},
        "G2": {"type": "boolean"},
        "G3": {"type": "boolean"},
        "G4": {"type": "boolean"}
      }
    },
    "metadata": {
      "type": "object",
      "properties": {
        "timestamp": {"type": "string", "format": "date-time"},
        "version": {"type": "string"},
        "degradation_level": {"type": "integer", "minimum": 0, "maximum": 4},
        "processing_time_seconds": {"type": "number"}
      }
    }
  },
  "required": ["language", "report_sections", "quality_gates", "metadata"]
}

Quality Gate Validation

Gate Enforcement Logic

For each gate in priority order:

1. Check: Validate gate condition
2. Auto-Fix: Attempt automatic remediation if failed
3. Retry: Re-check after auto-fix (max 2 retries)
4. Enforce: If still failed, apply enforcement action
5. Continue: Proceed to next gate or output limitation notice

Validation Schema

{
  "$schema": "http://json-schema.org/draft-07/schema#",
  "type": "object",
  "properties": {
    "gate_id": {"type": "string"},
    "status": {"type": "string", "enum": ["passed", "failed", "auto_fixed", "limitation_flagged"]},
    "attempts": {"type": "integer", "minimum": 1, "maximum": 3},
    "auto_fix_attempts": {"type": "integer", "minimum": 0, "maximum": 2},
    "enforcement_action": {"type": "string"},
    "timestamp": {"type": "string", "format": "date-time"}
  },
  "required": ["gate_id", "status", "attempts"]
}

Error Handling & Recovery

Error Categories

| Category | Examples | Recovery Strategy | |----------|----------|-------------------| | SOURCEUNAVAILABLE | Timeout, 429/500 errors | Retry with exponential backoff, fallback to secondary sources | | INVALIDINPUT | Schema validation failure | Request clarification, proceed with available fields | | MISSINGINPUT | Required field absent | Flag as DATAUNAVAILABLE, continue with degradation | | STALEDATA | Timestamp exceeds threshold | Flag with age, request refresh if critical | | KNOWLEDGEMISS | No matches in knowledge base | WebSearch gap-fill, queue for crawl | | CONFLICTINGACTIONS | Mutually exclusive actions | Apply precedence rules, document decision | | LIMITATIONFLAGGED | Quality gate cannot pass | Emit explicit limitation, continue degraded | | SYSTEM_ERROR | Unexpected exception | Log and telemetry, graceful degradation |

Error Recovery Process

[Error Detected] → [Categorize] → [Select Recovery] → [Execute Recovery] → [Verify] → [Continue/Retry/Fail]

Performance Monitoring

Metrics Tracked

• Execution Time: Time per sub-skill and total
• Token Usage: Input/output tokens per LLM call
• API Calls: Number and type of external API calls
• Cache Hit Rate: Percentage of cache hits
• Error Rate: Percentage of failed operations
• Quality Gate Pass Rate: Percentage of gates passed without auto-fix

Telemetry Schema

{
  "execution_id": "uuid",
  "timestamp": "iso8601",
  "metrics": {
    "execution_time_seconds": 0.0,
    "total_tokens": 0,
    "api_calls": 0,
    "cache_hit_rate": 0.0,
    "error_rate": 0.0,
    "quality_gate_pass_rate": 0.0
  },
  "performance": {
    "sub_skill_times": {},
    "tool_usage": {},
    "resource_utilization": {}
  }
}

Extension Points

Custom Sub-Skills

To add custom sub-skills:

1. Create sub-{name}.md in skills/ directory
2. Include frontmatter with name and description
3. Follow the standard sections: Role, Workflow, Tools, Output, Quality Gates
4. Register in main harness if required

Custom Tools

To add custom tools:

1. Define tool schema in tool registry
2. Implement execution handler
3. Add retry policy and error handling
4. Document usage examples

Custom Quality Gates

To add custom quality gates:

1. Define gate check criteria
2. Specify auto-fix procedure
3. Define enforcement logic
4. Add to quality gate table in main.md

Version Compatibility

• Claude Code: 1.0+
• Claude API: 1.0+
• Python: 3.8+
• Required Packages: See requirements.txt

License

MIT License - See LICENSE file for details.

Contributing

For contribution guidelines, see the project repository.

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/lead-free-heat-resistant-ceramics-agent-skill
• License: MIT

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