Pipeline run

{
  "final_skills": [
    {
      "is_primary": true,
      "skill_name": "AI"
    },
    {
      "is_primary": true,
      "skill_name": "Machine Learning"
    },
    {
      "is_primary": true,
      "skill_name": "Document Intelligence"
    },
    {
      "is_primary": true,
      "skill_name": "Retrieval"
    },
    {
      "is_primary": true,
      "skill_name": "Agentic Workflows"
    },
    {
      "is_primary": true,
      "skill_name": "Workflow Automation"
    }
  ],
  "jd_role": {
    "display_name": "Applied AI Engineer",
    "rationale": null,
    "role_archetype": "Engineering",
    "slug": ""
  },
  "nano_parsed": {
    "JD_type": "pass",
    "about_company": {
      "source_marker": {
        "first_5_words": "Bynd is building the intelligence",
        "last_5_words": "things that get used."
      },
      "text": "Bynd is building the intelligence layer for financial services.\nWe work with leading investment banks, private equity firms, asset managers, lenders, and advisory teams to transform how they extract, analyze, and act on information buried across financial documents, filings, reports, and internal workflows.\nWe operate with the standards of a research team and the urgency of a product company. We care deeply about technical quality, product taste, and building things that get used.",
      "word_count": 78
    },
    "certifications": [],
    "company_name": "Bynd",
    "ctc": null,
    "domain": {
      "primary": {
        "aliases": [
          "Investment Banking",
          "Private Equity",
          "Asset Management",
          "Lending",
          "Advisory"
        ],
        "domain": "Financial Services"
      },
      "secondary": null
    },
    "education": [],
    "experience": {
      "max": null,
      "min": null,
      "raw": null
    },
    "job_locations": [],
    "role": "Applied AI Engineer",
    "role_archetype": "Engineering",
    "roles_and_responsibilities": [
      {
        "bullet_count": 0,
        "heading": "Role Overview",
        "heading_was_present": false,
        "source_marker": {
          "first_5_words": "As an Applied AI Engineer",
          "last_5_words": "deploy them reliably in production."
        },
        "text": "As an Applied AI Engineer at Bynd, you will work across the core systems that power our product: document intelligence, retrieval, agentic workflows, and the infrastructure required to deploy them reliably in production.",
        "word_count": 31
      },
      {
        "bullet_count": 0,
        "heading": "Responsibilities",
        "heading_was_present": true,
        "source_marker": {
          "first_5_words": "Build systems that financial",
          "last_5_words": "workflow automation."
        },
        "text": "Build systems that financial institutions depend on for high-accuracy extraction, analysis, and workflow automation.",
        "word_count": 16
      }
    ],
    "urls": []
  },
  "run_id": null
}

{
  "alias_matches": [
    {
      "alias_persist_skipped_reason": "alias_text already exists for this canonical skill",
      "alias_persisted": false,
      "existing_alias_id": 3431,
      "existing_alias_text": "AI",
      "input_term": "AI",
      "matched_canonical": {
        "category_id": 2,
        "display_name": "AI",
        "id": 2634,
        "is_also_category": false,
        "is_extractable": true,
        "skill_nature": "CONCEPT",
        "slug": "ai",
        "sub_category_id": 2147,
        "typical_lifespan": "EVERGREEN",
        "volatility": "STABLE"
      },
      "matched_via": "alias"
    }
  ],
  "candidate_roles": [
    {
      "display_name": "ServiceNOW Developer",
      "id": 24,
      "rationale": null,
      "role_archetype": null,
      "slug": "servicenow-developer",
      "source": "db"
    }
  ],
  "chosen_role": {
    "display_name": "ServiceNOW Developer",
    "id": 24,
    "rationale": null,
    "role_archetype": null,
    "slug": "servicenow-developer",
    "source": "single_candidate"
  },
  "dimensions": [
    {
      "dimension": {
        "difficulty_hint": "well_known",
        "display_name": "Version Control Systems",
        "id": 365,
        "rationale": "Tools and workflows for tracking source changes, branching, merging, and collaborating on code history. Git belongs here because it is the canonical distributed version control system used to manage revisions and coordinate team development.",
        "slug": "d_init_01",
        "source": "db"
      },
      "input_skill": "AI",
      "llm_role": null,
      "roles_from_db": []
    },
    {
      "dimension": {
        "difficulty_hint": "well_known",
        "display_name": "Version Control Systems",
        "id": 365,
        "rationale": "Tools and workflows for tracking source changes, branching, merging, and collaborating on code history. Git belongs here because it is the canonical distributed version control system used to manage revisions and coordinate team development.",
        "slug": "d_init_01",
        "source": "db"
      },
      "input_skill": "Machine Learning",
      "llm_role": null,
      "roles_from_db": []
    },
    {
      "dimension": {
        "difficulty_hint": "well_known",
        "display_name": "Project Delivery and Coordination",
        "id": 366,
        "rationale": "Coordination practices for organizing work, tracking progress, and aligning stakeholders across a delivery effort. Agile fits here when used as a team execution framework for managing scope, cadence, and collaboration.",
        "slug": "d_init_02",
        "source": "db"
      },
      "input_skill": "Machine Learning",
      "llm_role": null,
      "roles_from_db": []
    },
    {
      "dimension": {
        "difficulty_hint": "well_known",
        "display_name": "Version Control Systems",
        "id": 365,
        "rationale": "Tools and workflows for tracking source changes, branching, merging, and collaborating on code history. Git belongs here because it is the canonical distributed version control system used to manage revisions and coordinate team development.",
        "slug": "d_init_01",
        "source": "db"
      },
      "input_skill": "Document Intelligence",
      "llm_role": null,
      "roles_from_db": []
    },
    {
      "dimension": {
        "difficulty_hint": null,
        "display_name": "Context Management and Retrieval for LLM Calls",
        "id": null,
        "rationale": "Preparing, selecting, summarizing, and packaging information for model calls so responses stay relevant and grounded. Includes retrieving documents, passages, or structured records; chunking and ranking results; assembling prompt context; and using retrieval-augmented generation, vector search, hybrid search, reranking, and citation grounding to decide what information is included at call time.",
        "slug": "d_merge_01",
        "source": "llm"
      },
      "input_skill": "Retrieval",
      "llm_role": null,
      "roles_from_db": []
    },
    {
      "dimension": {
        "difficulty_hint": null,
        "display_name": "Context Management and Retrieval for AI Systems",
        "id": null,
        "rationale": "Preparing, selecting, summarizing, packaging, and retrieving the right context for AI model or agent calls so outputs stay relevant, grounded, and accurate across steps. This includes context-window management, retrieval-augmented generation, memory selection, grounding data, conversation state, tool-result packaging, prompt context assembly, and other mechanisms for passing the right information into model interactions or agentic workflows.",
        "slug": "d_merge_01",
        "source": "llm"
      },
      "input_skill": "Agentic Workflows",
      "llm_role": null,
      "roles_from_db": []
    },
    {
      "dimension": {
        "difficulty_hint": "well_known",
        "display_name": "Workflow Automation and Approvals",
        "id": 211,
        "rationale": "Designing and configuring workflow-driven process automation, including approvals, task routing, and lifecycle transitions. This cluster is coherent because ServiceNow developers often implement process logic rather than standalone application code.",
        "slug": "workflow-automation-and-approvals",
        "source": "db"
      },
      "input_skill": "Agentic Workflows",
      "llm_role": null,
      "roles_from_db": [
        {
          "display_name": "ServiceNOW Developer",
          "id": 24,
          "rationale": null,
          "role_archetype": null,
          "slug": "servicenow-developer",
          "source": "db"
        }
      ]
    },
    {
      "dimension": {
        "difficulty_hint": "well_known",
        "display_name": "Workflow Automation and Approvals",
        "id": 211,
        "rationale": "Designing and configuring workflow-driven process automation, including approvals, task routing, and lifecycle transitions. This cluster is coherent because ServiceNow developers often implement process logic rather than standalone application code.",
        "slug": "workflow-automation-and-approvals",
        "source": "db"
      },
      "input_skill": "Agentic Workflows",
      "llm_role": null,
      "roles_from_db": [
        {
          "display_name": "ServiceNOW Developer",
          "id": 24,
          "rationale": null,
          "role_archetype": null,
          "slug": "servicenow-developer",
          "source": "db"
        }
      ]
    },
    {
      "dimension": {
        "difficulty_hint": "well_known",
        "display_name": "Workflow Automation and Approvals",
        "id": 211,
        "rationale": "Designing and configuring workflow-driven process automation, including approvals, task routing, and lifecycle transitions. This cluster is coherent because ServiceNow developers often implement process logic rather than standalone application code.",
        "slug": "workflow-automation-and-approvals",
        "source": "db"
      },
      "input_skill": "Workflow Automation",
      "llm_role": null,
      "roles_from_db": [
        {
          "display_name": "ServiceNOW Developer",
          "id": 24,
          "rationale": null,
          "role_archetype": null,
          "slug": "servicenow-developer",
          "source": "db"
        }
      ]
    },
    {
      "dimension": {
        "difficulty_hint": "well_known",
        "display_name": "Workflow Automation and Approvals",
        "id": 211,
        "rationale": "Designing and configuring workflow-driven process automation, including approvals, task routing, and lifecycle transitions. This cluster is coherent because ServiceNow developers often implement process logic rather than standalone application code.",
        "slug": "workflow-automation-and-approvals",
        "source": "db"
      },
      "input_skill": "Workflow Automation",
      "llm_role": null,
      "roles_from_db": [
        {
          "display_name": "ServiceNOW Developer",
          "id": 24,
          "rationale": null,
          "role_archetype": null,
          "slug": "servicenow-developer",
          "source": "db"
        }
      ]
    }
  ],
  "input_final_skills": [
    "AI",
    "Machine Learning",
    "Document Intelligence",
    "Retrieval",
    "Agentic Workflows",
    "Workflow Automation"
  ],
  "input_llm_skills": [
    "AI",
    "Machine Learning",
    "Document Intelligence",
    "Retrieval",
    "Agentic Workflows",
    "Workflow Automation"
  ],
  "new_aliases_persisted": 0,
  "run_id": "cb714ab7-756c-437d-ac85-e5b6557aefa5",
  "skills_detail": [
    {
      "aliases_in_db": [
        {
          "alias_text": "AI",
          "alias_type": "CANONICAL",
          "id": 3431,
          "is_primary": false,
          "match_strategy": "CASE_INSENSITIVE"
        }
      ],
      "canonical": {
        "category_id": 2,
        "display_name": "AI",
        "id": 2634,
        "is_also_category": false,
        "is_extractable": true,
        "skill_nature": "CONCEPT",
        "slug": "ai",
        "sub_category_id": 2147,
        "typical_lifespan": "EVERGREEN",
        "volatility": "STABLE"
      },
      "dimensions": [
        {
          "dimension": {
            "difficulty_hint": "well_known",
            "display_name": "Version Control Systems",
            "id": 365,
            "rationale": "Tools and workflows for tracking source changes, branching, merging, and collaborating on code history. Git belongs here because it is the canonical distributed version control system used to manage revisions and coordinate team development.",
            "slug": "d_init_01",
            "source": "db"
          },
          "input_skill": "AI",
          "llm_role": null,
          "roles_from_db": []
        }
      ],
      "input_skill": "AI",
      "matched_via": "alias",
      "new_alias_persisted": false,
      "new_alias_text": null,
      "new_skill_meta": null,
      "source_tag": "db",
      "was_in_llm_skills": true
    },
    {
      "aliases_in_db": [],
      "canonical": null,
      "dimensions": [
        {
          "dimension": {
            "difficulty_hint": "well_known",
            "display_name": "Version Control Systems",
            "id": 365,
            "rationale": "Tools and workflows for tracking source changes, branching, merging, and collaborating on code history. Git belongs here because it is the canonical distributed version control system used to manage revisions and coordinate team development.",
            "slug": "d_init_01",
            "source": "db"
          },
          "input_skill": "Machine Learning",
          "llm_role": null,
          "roles_from_db": []
        },
        {
          "dimension": {
            "difficulty_hint": "well_known",
            "display_name": "Project Delivery and Coordination",
            "id": 366,
            "rationale": "Coordination practices for organizing work, tracking progress, and aligning stakeholders across a delivery effort. Agile fits here when used as a team execution framework for managing scope, cadence, and collaboration.",
            "slug": "d_init_02",
            "source": "db"
          },
          "input_skill": "Machine Learning",
          "llm_role": null,
          "roles_from_db": []
        }
      ],
      "input_skill": "Machine Learning",
      "matched_via": null,
      "new_alias_persisted": false,
      "new_alias_text": null,
      "new_skill_meta": {
        "derived": {
          "category": "Concept",
          "skill_nature": "CONCEPT",
          "sub_category": "machine_learning",
          "typical_lifespan": "EVERGREEN",
          "version_strategy": "NOT_APPLICABLE",
          "volatility": "STABLE"
        },
        "enrichment": {
          "ambiguity": {
            "ambiguity_flag": false,
            "confused_with": [],
            "reasoning": "\u201cMachine Learning\u201d is a standard, well-defined concept name in JDs and is unlikely to be mistaken for a different catalog skill in typical context."
          },
          "context_keywords": {
            "context_keywords": [
              "scikit-learn",
              "TensorFlow",
              "PyTorch",
              "XGBoost",
              "feature engineering",
              "model training",
              "cross-validation",
              "hyperparameter tuning",
              "classification",
              "regression",
              "clustering",
              "gradient descent",
              "random forest",
              "neural networks",
              "MLOps"
            ]
          },
          "maturity": {
            "confidence": 0.97,
            "maturity": "well_known",
            "reasoning": "Machine Learning appears in a large share of software, data, and AI job descriptions, and major cloud vendors (AWS, Google Cloud, Azure) offer dedicated ML platforms, indicating broad hiring demand and adoption."
          },
          "skill_id": "machine-learning",
          "vendor_license": {
            "confidence": 0.99,
            "license": null,
            "vendor": null,
            "year_introduced": null
          },
          "versioning": {
            "current_version": null,
            "version_aliases": {},
            "versioned": false
          }
        },
        "keep_log": [],
        "locked_dimensions": [
          {
            "description": "Core methods for building predictive and adaptive models from data. This covers the modeling concepts, training workflows, and evaluation practices that make Machine Learning a distinct engineering skill.",
            "exemplar_skills": [
              "Machine Learning",
              "supervised learning",
              "unsupervised learning",
              "feature engineering",
              "model training",
              "cross-validation",
              "classification",
              "regression"
            ],
            "in_scope": "Machine Learning, supervised learning, unsupervised learning, feature engineering, model training, model selection, cross-validation, classification, regression, clustering, dimensionality reduction",
            "name": "Machine Learning Methods",
            "out_of_scope": "Model deployment, inference packaging, and serving infrastructure; those belong to model-serving-deployment-and-runtime-packaging and model-serving-architecture. Experiment design and causal analysis belong to experiment-design-and-analysis.",
            "overlap_flags": [
              {
                "reason": "ML work often uses experiments for evaluation, but the core dimension here is model building rather than causal testing.",
                "with_dim_id": "experiment-design-and-analysis",
                "with_dim_name": null,
                "with_role": "Data Scientist"
              },
              {
                "reason": "Trained ML models are frequently deployed, but serving and packaging are operational concerns separate from the modeling skill itself.",
                "with_dim_id": "model-serving-deployment-and-runtime-packaging",
                "with_dim_name": null,
                "with_role": "MLOps Engineer, Machine Learning Engineer"
              }
            ],
            "tentative_id": "d_init_01"
          },
          {
            "description": "Using machine learning techniques to solve real product and business problems with measurable outcomes. This fits an Applied AI Engineer because the skill is typically exercised in building practical predictive systems rather than purely theoretical ML research.",
            "exemplar_skills": [
              "Machine Learning",
              "predictive modeling",
              "model evaluation",
              "feature selection",
              "model tuning",
              "anomaly detection",
              "recommendation systems",
              "training data preparation"
            ],
            "in_scope": "Machine Learning, predictive modeling, model evaluation, training data preparation, feature selection, model tuning, classification, regression, anomaly detection, recommendation modeling",
            "name": "Applied Predictive Modeling",
            "out_of_scope": "General software integration, API wiring, and runtime orchestration belong to application or serving dimensions. Statistical experiment design and impact estimation belong to experiment-design-and-analysis.",
            "overlap_flags": [
              {
                "reason": "Applied ML often relies on A/B tests and offline metrics, but those are evaluation methods rather than the predictive modeling dimension itself.",
                "with_dim_id": "experiment-design-and-analysis",
                "with_dim_name": null,
                "with_role": "Data Scientist"
              },
              {
                "reason": "Some applied AI systems combine ML with retrieval, but retrieval-specific context packaging is a separate capability.",
                "with_dim_id": "context-management-and-retrieval",
                "with_dim_name": null,
                "with_role": "AI Engineer"
              }
            ],
            "tentative_id": "d_init_02"
          }
        ],
        "merge_log": [],
        "placed": {
          "name": "Machine Learning",
          "placement_confidence": 0.92,
          "primary_dimension": "d_init_01",
          "reasoning": "Deterministic JD placement: locked_dimensions has 2 dimension(s) from skill-driven dimension generation after reconciliation; primary_dimension is the first locked dim.",
          "secondary_dimensions": [
            "d_init_02"
          ],
          "skill_id": "machine-learning"
        },
        "relationships": {
          "child_skills": [],
          "parent_skills": [
            "ai-ml",
            "ai"
          ],
          "related_to": [
            "algorithms",
            "mlops",
            "mlflow",
            "amazon-sagemaker",
            "hybrid-retrieval",
            "python",
            "devops",
            "llm",
            "vision-language-models",
            "multimodal-document-understanding"
          ],
          "requires": [],
          "skill_id": "machine-learning",
          "suppress_on_match": []
        },
        "skill_id": "machine-learning",
        "split_log": [],
        "typed": {
          "alternatives_considered": [],
          "confidence": 0.98,
          "name": "Machine Learning",
          "reasoning": "Machine Learning is fundamentally a named knowledge unit about building systems that learn from data, so it fits the Concept type rather than a Methodology or Architecture.",
          "skill_id": "machine-learning",
          "subtype": "machine_learning",
          "type": "Concept"
        },
        "warnings": []
      },
      "source_tag": "llm",
      "was_in_llm_skills": true
    },
    {
      "aliases_in_db": [],
      "canonical": null,
      "dimensions": [
        {
          "dimension": {
            "difficulty_hint": "well_known",
            "display_name": "Version Control Systems",
            "id": 365,
            "rationale": "Tools and workflows for tracking source changes, branching, merging, and collaborating on code history. Git belongs here because it is the canonical distributed version control system used to manage revisions and coordinate team development.",
            "slug": "d_init_01",
            "source": "db"
          },
          "input_skill": "Document Intelligence",
          "llm_role": null,
          "roles_from_db": []
        }
      ],
      "input_skill": "Document Intelligence",
      "matched_via": null,
      "new_alias_persisted": false,
      "new_alias_text": null,
      "new_skill_meta": {
        "derived": {
          "category": "Platform",
          "skill_nature": "PLATFORM",
          "sub_category": "document_intelligence_platform",
          "typical_lifespan": "EVERGREEN",
          "version_strategy": "NOT_APPLICABLE",
          "volatility": "EMERGING"
        },
        "enrichment": {
          "ambiguity": {
            "ambiguity_flag": false,
            "confused_with": [],
            "reasoning": "\u201cDocument Intelligence\u201d is a fairly specific platform name and is unlikely to be mistaken for a different catalog skill in typical job descriptions."
          },
          "context_keywords": {
            "context_keywords": [
              "OCR",
              "layout analysis",
              "document classification",
              "entity extraction",
              "form processing",
              "invoice processing",
              "receipt capture",
              "table extraction",
              "key-value pairs",
              "PDF parsing",
              "handwriting recognition",
              "semantic segmentation",
              "document understanding",
              "text extraction",
              "template matching"
            ]
          },
          "maturity": {
            "confidence": 0.78,
            "maturity": "emerging",
            "reasoning": "Document intelligence platforms are increasingly appearing in enterprise JDs for OCR, extraction, and workflow automation, but the market is still fragmented across Azure/AWS/Google offerings rather than a universal staple."
          },
          "skill_id": "document-intelligence",
          "vendor_license": {
            "confidence": 0.93,
            "license": "proprietary",
            "vendor": "Microsoft",
            "year_introduced": 2018
          },
          "versioning": {
            "current_version": null,
            "version_aliases": {},
            "versioned": false
          }
        },
        "keep_log": [],
        "locked_dimensions": [
          {
            "description": "Techniques and systems for extracting, classifying, and structuring information from documents such as PDFs, scans, forms, and images. This skill belongs here because it centers on understanding document content with OCR, layout analysis, and information extraction workflows.",
            "exemplar_skills": [
              "Document Intelligence",
              "OCR",
              "layout analysis",
              "form extraction",
              "table extraction",
              "key-value extraction",
              "document classification",
              "invoice extraction",
              "receipt parsing"
            ],
            "in_scope": "Document Intelligence, OCR, layout analysis, form extraction, table extraction, key-value extraction, document classification, document parsing, invoice/receipt extraction, contract understanding, scanned PDF processing, handwriting recognition, document AI pipelines",
            "name": "Document Intelligence",
            "out_of_scope": "general computer vision object detection, chatbot prompt design, model serving infrastructure, database query optimization, UI document viewers",
            "overlap_flags": [
              {
                "reason": "Document intelligence systems often feed extracted text into retrieval pipelines, but this dimension focuses on document understanding rather than context packaging for LLM calls.",
                "with_dim_id": "context-management-and-retrieval",
                "with_dim_name": null,
                "with_role": "AI Engineer"
              },
              {
                "reason": "Document processing may be operationalized as pipelines, but that dimension owns the movement and preparation of data rather than the extraction semantics.",
                "with_dim_id": "inference-data-pipelines",
                "with_dim_name": null,
                "with_role": "MLOps Engineer"
              }
            ],
            "tentative_id": "d_init_01"
          }
        ],
        "merge_log": [],
        "placed": {
          "name": "Document Intelligence",
          "placement_confidence": 0.92,
          "primary_dimension": "d_init_01",
          "reasoning": "Deterministic JD placement: locked_dimensions has 1 dimension(s) from skill-driven dimension generation after reconciliation; primary_dimension is the first locked dim.",
          "secondary_dimensions": [],
          "skill_id": "document-intelligence"
        },
        "relationships": {
          "child_skills": [],
          "parent_skills": [],
          "related_to": [
            "multimodal-document-understanding",
            "ocr",
            "ai-ml",
            "ai",
            "digital-signatures",
            "evidence-preservation",
            "hybrid-retrieval",
            "agentic-systems",
            "elasticsearch",
            "observability"
          ],
          "requires": [],
          "skill_id": "document-intelligence",
          "suppress_on_match": []
        },
        "skill_id": "document-intelligence",
        "split_log": [],
        "typed": {
          "alternatives_considered": [
            "Service: ruled out \u2014 the name refers to the overall vendor-hosted product rather than a single capability inside a larger platform.",
            "Tool: ruled out \u2014 this is not primarily self-hosted software operated by the user."
          ],
          "confidence": 0.88,
          "name": "Document Intelligence",
          "reasoning": "By the Vendor SaaS = Platform rule, Document Intelligence is best treated as a hosted multi-tenant managed offering rather than software you run yourself.",
          "skill_id": "document-intelligence",
          "subtype": "document_intelligence_platform",
          "type": "Platform"
        },
        "warnings": []
      },
      "source_tag": "llm",
      "was_in_llm_skills": true
    },
    {
      "aliases_in_db": [],
      "canonical": null,
      "dimensions": [
        {
          "dimension": {
            "difficulty_hint": null,
            "display_name": "Context Management and Retrieval for LLM Calls",
            "id": null,
            "rationale": "Preparing, selecting, summarizing, and packaging information for model calls so responses stay relevant and grounded. Includes retrieving documents, passages, or structured records; chunking and ranking results; assembling prompt context; and using retrieval-augmented generation, vector search, hybrid search, reranking, and citation grounding to decide what information is included at call time.",
            "slug": "d_merge_01",
            "source": "llm"
          },
          "input_skill": "Retrieval",
          "llm_role": null,
          "roles_from_db": []
        }
      ],
      "input_skill": "Retrieval",
      "matched_via": null,
      "new_alias_persisted": false,
      "new_alias_text": null,
      "new_skill_meta": {
        "derived": {
          "category": "Concept",
          "skill_nature": "CONCEPT",
          "sub_category": "information_retrieval",
          "typical_lifespan": "EVERGREEN",
          "version_strategy": "NOT_APPLICABLE",
          "volatility": "STABLE"
        },
        "enrichment": {
          "ambiguity": {
            "ambiguity_flag": false,
            "confused_with": [],
            "reasoning": "\u201cRetrieval\u201d in a JD typically refers to information retrieval/search systems, which is a distinct concept and not a common shorthand for another catalog skill."
          },
          "context_keywords": {
            "context_keywords": [
              "BM25",
              "TF-IDF",
              "vector search",
              "embeddings",
              "semantic search",
              "ranker",
              "relevance ranking",
              "query expansion",
              "indexing",
              "Lucene",
              "Elasticsearch",
              "Apache Solr",
              "ANN",
              "hybrid search",
              "reranking"
            ]
          },
          "maturity": {
            "confidence": 0.86,
            "maturity": "well_known",
            "reasoning": "Broadly used across search, recommender, and RAG roles; job postings commonly list retrieval/search relevance alongside Elasticsearch/OpenSearch and vector DBs, indicating strong hiring demand."
          },
          "skill_id": "retrieval",
          "vendor_license": {
            "confidence": 0.99,
            "license": null,
            "vendor": null,
            "year_introduced": null
          },
          "versioning": {
            "current_version": null,
            "version_aliases": {},
            "versioned": false
          }
        },
        "keep_log": [],
        "locked_dimensions": [
          {
            "description": "Preparing, selecting, summarizing, and packaging information for model calls so responses stay relevant and grounded. Includes retrieving documents, passages, or structured records; chunking and ranking results; assembling prompt context; and using retrieval-augmented generation, vector search, hybrid search, reranking, and citation grounding to decide what information is included at call time.",
            "exemplar_skills": [
              "Context Management and Retrieval for LLM Calls"
            ],
            "in_scope": "Skills, tools, and practices that belong under Context Management and Retrieval for LLM Calls for the target role, including items implied by the dimension rationale.",
            "name": "Context Management and Retrieval for LLM Calls",
            "out_of_scope": "Adjacent clusters explicitly not owned by Context Management and Retrieval for LLM Calls, including unrelated platforms, roles, and skill families per library policy.",
            "overlap_flags": [],
            "tentative_id": "d_merge_01"
          }
        ],
        "merge_log": [
          {
            "a_dim_id": "context-management-and-retrieval",
            "a_name": "Context Management and Retrieval",
            "a_role": "__skill_focal__",
            "b_dim_id": "context-management-and-retrieval",
            "b_name": "Context Management and Retrieval",
            "b_role": "AI Engineer",
            "into": "d_merge_01",
            "into_name": "Context Management and Retrieval for LLM Calls",
            "merged_from": [
              "context-management-and-retrieval",
              "context-management-and-retrieval"
            ],
            "pair_kind": "cross_role",
            "reasoning": "Both dims define the same cluster: preparing and selecting context for model calls. Dim A covers retrieval, chunking, top-k retrieval, reranking, vector/hybrid search, RAG, and citation grounding; Dim B repeats the same core description and says AI features depend on what is included or retrieved at call time. B has no distinct exemplars or scope, so this is not a separate role-specific cluster.",
            "similarity": 0.8904107828261284
          }
        ],
        "placed": {
          "name": "Retrieval",
          "placement_confidence": 0.92,
          "primary_dimension": "d_merge_01",
          "reasoning": "Deterministic JD placement: locked_dimensions has 1 dimension(s) from skill-driven dimension generation after reconciliation; primary_dimension is the first locked dim.",
          "secondary_dimensions": [],
          "skill_id": "retrieval"
        },
        "relationships": {
          "child_skills": [],
          "parent_skills": [],
          "related_to": [
            "hybrid-retrieval",
            "evaluation",
            "reranking",
            "ocr",
            "context-management",
            "rag",
            "sql",
            "prompting",
            "authentication"
          ],
          "requires": [],
          "skill_id": "retrieval",
          "suppress_on_match": []
        },
        "skill_id": "retrieval",
        "split_log": [],
        "typed": {
          "alternatives_considered": [],
          "confidence": 0.88,
          "name": "Retrieval",
          "reasoning": "Retrieval is fundamentally a knowledge unit about finding and obtaining information, so it fits the Concept type rather than a tool, platform, or methodology.",
          "skill_id": "retrieval",
          "subtype": "information_retrieval",
          "type": "Concept"
        },
        "warnings": [
          "stage3_post_filter_dropped_catalog_only_locked_dims:40-\u003e1"
        ]
      },
      "source_tag": "llm",
      "was_in_llm_skills": true
    },
    {
      "aliases_in_db": [],
      "canonical": null,
      "dimensions": [
        {
          "dimension": {
            "difficulty_hint": null,
            "display_name": "Context Management and Retrieval for AI Systems",
            "id": null,
            "rationale": "Preparing, selecting, summarizing, packaging, and retrieving the right context for AI model or agent calls so outputs stay relevant, grounded, and accurate across steps. This includes context-window management, retrieval-augmented generation, memory selection, grounding data, conversation state, tool-result packaging, prompt context assembly, and other mechanisms for passing the right information into model interactions or agentic workflows.",
            "slug": "d_merge_01",
            "source": "llm"
          },
          "input_skill": "Agentic Workflows",
          "llm_role": null,
          "roles_from_db": []
        },
        {
          "dimension": {
            "difficulty_hint": "well_known",
            "display_name": "Workflow Automation and Approvals",
            "id": 211,
            "rationale": "Designing and configuring workflow-driven process automation, including approvals, task routing, and lifecycle transitions. This cluster is coherent because ServiceNow developers often implement process logic rather than standalone application code.",
            "slug": "workflow-automation-and-approvals",
            "source": "db"
          },
          "input_skill": "Agentic Workflows",
          "llm_role": null,
          "roles_from_db": [
            {
              "display_name": "ServiceNOW Developer",
              "id": 24,
              "rationale": null,
              "role_archetype": null,
              "slug": "servicenow-developer",
              "source": "db"
            }
          ]
        },
        {
          "dimension": {
            "difficulty_hint": "well_known",
            "display_name": "Workflow Automation and Approvals",
            "id": 211,
            "rationale": "Designing and configuring workflow-driven process automation, including approvals, task routing, and lifecycle transitions. This cluster is coherent because ServiceNow developers often implement process logic rather than standalone application code.",
            "slug": "workflow-automation-and-approvals",
            "source": "db"
          },
          "input_skill": "Agentic Workflows",
          "llm_role": null,
          "roles_from_db": [
            {
              "display_name": "ServiceNOW Developer",
              "id": 24,
              "rationale": null,
              "role_archetype": null,
              "slug": "servicenow-developer",
              "source": "db"
            }
          ]
        }
      ],
      "input_skill": "Agentic Workflows",
      "matched_via": null,
      "new_alias_persisted": false,
      "new_alias_text": null,
      "new_skill_meta": {
        "derived": {
          "category": "Architecture",
          "skill_nature": "PATTERN",
          "sub_category": "agentic_workflows",
          "typical_lifespan": "EVERGREEN",
          "version_strategy": "NOT_APPLICABLE",
          "volatility": "EMERGING"
        },
        "enrichment": {
          "ambiguity": {
            "ambiguity_flag": false,
            "confused_with": [],
            "reasoning": "\u201cAgentic Workflows\u201d is a fairly specific AI/automation architecture term and is unlikely to be mistaken for another catalog skill in a typical JD."
          },
          "context_keywords": {
            "context_keywords": [
              "LLM orchestration",
              "tool calling",
              "function calling",
              "planner-executor",
              "ReAct",
              "multi-agent",
              "workflow orchestration",
              "state machine",
              "task decomposition",
              "retrieval-augmented generation",
              "memory management",
              "prompt chaining",
              "autonomous agents",
              "guardrails",
              "human-in-the-loop"
            ]
          },
          "maturity": {
            "confidence": 0.84,
            "maturity": "emerging",
            "reasoning": "Job postings increasingly mention agentic workflows alongside LLM orchestration, but JD volume is still far below core stacks like Python/AWS; market demand is growing, not yet universal."
          },
          "skill_id": "agentic-workflows",
          "vendor_license": {
            "confidence": 0.98,
            "license": null,
            "vendor": null,
            "year_introduced": null
          },
          "versioning": {
            "current_version": null,
            "version_aliases": {},
            "versioned": false
          }
        },
        "keep_log": [
          {
            "a_dim_id": "workflow-automation-and-approvals",
            "a_name": "Workflow Automation and Approvals",
            "a_role": "__skill_focal__",
            "b_dim_id": "workflow-automation-and-approvals",
            "b_name": "Workflow Automation and Approvals",
            "b_role": "ServiceNOW Developer",
            "pair_kind": "cross_role",
            "reasoning": "Dim A is platform-agnostic workflow orchestration: agentic workflows, conditional branching, human-in-the-loop steps, and tools like Power Automate. Dim B is ServiceNow-specific process automation for developers, focused on approvals, task routing, and lifecycle transitions inside ServiceNow. The overlap is generic workflow/approval language, but the concrete skills and implementation context differ, so they are not the same cluster.",
            "similarity": 0.7603933656935488
          }
        ],
        "locked_dimensions": [
          {
            "description": "Preparing, selecting, summarizing, packaging, and retrieving the right context for AI model or agent calls so outputs stay relevant, grounded, and accurate across steps. This includes context-window management, retrieval-augmented generation, memory selection, grounding data, conversation state, tool-result packaging, prompt context assembly, and other mechanisms for passing the right information into model interactions or agentic workflows.",
            "exemplar_skills": [
              "Context Management and Retrieval for AI Systems"
            ],
            "in_scope": "Skills, tools, and practices that belong under Context Management and Retrieval for AI Systems for the target role, including items implied by the dimension rationale.",
            "name": "Context Management and Retrieval for AI Systems",
            "out_of_scope": "Adjacent clusters explicitly not owned by Context Management and Retrieval for AI Systems, including unrelated platforms, roles, and skill families per library policy.",
            "overlap_flags": [],
            "tentative_id": "d_merge_01"
          },
          {
            "description": "Designing and configuring multi-step workflows that route work, trigger actions, and coordinate approvals or handoffs. Agentic workflows fit here when they orchestrate tasks across tools and decision points rather than just single-step automation.",
            "exemplar_skills": [
              "Agentic Workflows",
              "workflow orchestration",
              "approval routing",
              "task handoffs",
              "conditional branching",
              "human-in-the-loop automation",
              "process automation"
            ],
            "in_scope": "Agentic Workflows, workflow orchestration, approval routing, task handoffs, conditional branching, human-in-the-loop steps, lifecycle transitions, process automation, Power Automate, ServiceNow workflows, orchestration rules",
            "name": "Workflow Automation and Approvals",
            "out_of_scope": "Model training and prompt optimization, which belong to AI model development; low-level API integration and data fetching, which belong to integration patterns; UI component behavior, which belongs to frontend dimensions",
            "overlap_flags": [
              {
                "reason": "Agentic workflows often call external services and APIs, so integration boundaries can overlap with orchestration design.",
                "with_dim_id": "cloud-service-integration-patterns",
                "with_dim_name": null,
                "with_role": "Cloud Architect"
              },
              {
                "reason": "Event-driven workflow steps may use queues or streams to trigger downstream actions asynchronously.",
                "with_dim_id": "messaging-and-event-streaming",
                "with_dim_name": null,
                "with_role": "Backend Engineer"
              }
            ],
            "tentative_id": "workflow-automation-and-approvals"
          },
          {
            "description": "Designing and configuring workflow-driven process automation, including approvals, task routing, and lifecycle transitions. This cluster is coherent because ServiceNow developers often implement process logic rather than standalone application code.",
            "exemplar_skills": [
              "Workflow Automation and Approvals"
            ],
            "in_scope": "Skills, tools, and practices that belong under Workflow Automation and Approvals for the target role, including items implied by the dimension rationale.",
            "name": "Workflow Automation and Approvals",
            "out_of_scope": "Adjacent clusters explicitly not owned by Workflow Automation and Approvals, including unrelated platforms, roles, and skill families per library policy.",
            "overlap_flags": [],
            "tentative_id": "workflow-automation-and-approvals"
          }
        ],
        "merge_log": [
          {
            "a_dim_id": "context-management-and-retrieval",
            "a_name": "Context Management and Retrieval",
            "a_role": "__skill_focal__",
            "b_dim_id": "context-management-and-retrieval",
            "b_name": "Context Management and Retrieval",
            "b_role": "AI Engineer",
            "into": "d_merge_01",
            "into_name": "Context Management and Retrieval for AI Systems",
            "merged_from": [
              "context-management-and-retrieval",
              "context-management-and-retrieval"
            ],
            "pair_kind": "cross_role",
            "reasoning": "These two dimensions describe the same conceptual cluster with nearly identical wording and substance. Dim A explicitly covers \"Preparing, selecting, and retrieving the right context for AI agents\" and lists exemplar skills like Agentic Workflows, retrieval-augmented generation, memory selection, tool-result packaging, and prompt context assembly. Dim B says \"Preparing, selecting, and packaging context for model calls so responses stay relevant and grounded\" and frames the same core activity at call time. The overlap is not just lexical: both center on deciding what information enters an AI interaction, how it is retrieved/assembled, and how it is grounded. The difference is only scope phrasing (agents across steps vs model calls), not a distinct skill cluster.",
            "similarity": 0.8232155368004
          }
        ],
        "placed": {
          "name": "Agentic Workflows",
          "placement_confidence": 0.92,
          "primary_dimension": "d_merge_01",
          "reasoning": "Deterministic JD placement: locked_dimensions has 3 dimension(s) from skill-driven dimension generation after reconciliation; primary_dimension is the first locked dim.",
          "secondary_dimensions": [
            "workflow-automation-and-approvals"
          ],
          "skill_id": "agentic-workflows"
        },
        "relationships": {
          "child_skills": [],
          "parent_skills": [],
          "related_to": [
            "agentic-systems",
            "context-management",
            "structured-outputs",
            "prompting",
            "devops",
            "ci-cd",
            "mlops",
            "ai-ml",
            "rest-apis",
            "amazon-api-gateway"
          ],
          "requires": [],
          "skill_id": "agentic-workflows",
          "suppress_on_match": []
        },
        "skill_id": "agentic-workflows",
        "split_log": [],
        "typed": {
          "alternatives_considered": [
            "Concept: ruled out \u2014 it is more than a single knowledge unit and instead describes an overall build pattern.",
            "Methodology: ruled out \u2014 it is not primarily a process or way of working like Agile or TDD."
          ],
          "confidence": 0.82,
          "name": "Agentic Workflows",
          "reasoning": "Agentic Workflows is best treated as an Architecture because it describes a system-shape/pattern for how autonomous agents are organized and interact, rather than a specific tool, language, or methodology.",
          "skill_id": "agentic-workflows",
          "subtype": "agentic_workflows",
          "type": "Architecture"
        },
        "warnings": [
          "stage3_post_filter_dropped_catalog_only_locked_dims:41-\u003e3"
        ]
      },
      "source_tag": "llm",
      "was_in_llm_skills": true
    },
    {
      "aliases_in_db": [],
      "canonical": null,
      "dimensions": [
        {
          "dimension": {
            "difficulty_hint": "well_known",
            "display_name": "Workflow Automation and Approvals",
            "id": 211,
            "rationale": "Designing and configuring workflow-driven process automation, including approvals, task routing, and lifecycle transitions. This cluster is coherent because ServiceNow developers often implement process logic rather than standalone application code.",
            "slug": "workflow-automation-and-approvals",
            "source": "db"
          },
          "input_skill": "Workflow Automation",
          "llm_role": null,
          "roles_from_db": [
            {
              "display_name": "ServiceNOW Developer",
              "id": 24,
              "rationale": null,
              "role_archetype": null,
              "slug": "servicenow-developer",
              "source": "db"
            }
          ]
        },
        {
          "dimension": {
            "difficulty_hint": "well_known",
            "display_name": "Workflow Automation and Approvals",
            "id": 211,
            "rationale": "Designing and configuring workflow-driven process automation, including approvals, task routing, and lifecycle transitions. This cluster is coherent because ServiceNow developers often implement process logic rather than standalone application code.",
            "slug": "workflow-automation-and-approvals",
            "source": "db"
          },
          "input_skill": "Workflow Automation",
          "llm_role": null,
          "roles_from_db": [
            {
              "display_name": "ServiceNOW Developer",
              "id": 24,
              "rationale": null,
              "role_archetype": null,
              "slug": "servicenow-developer",
              "source": "db"
            }
          ]
        }
      ],
      "input_skill": "Workflow Automation",
      "matched_via": null,
      "new_alias_persisted": false,
      "new_alias_text": null,
      "new_skill_meta": {
        "derived": {
          "category": "Methodology",
          "skill_nature": "METHODOLOGY",
          "sub_category": "workflow_automation",
          "typical_lifespan": "EVERGREEN",
          "version_strategy": "NOT_APPLICABLE",
          "volatility": "STABLE"
        },
        "enrichment": {
          "ambiguity": {
            "ambiguity_flag": false,
            "confused_with": [],
            "reasoning": "The term is broad but usually refers to automating business/process workflows, not a distinct overloaded product or acronym. In typical JDs it is unlikely to be mistaken for another catalog skill."
          },
          "context_keywords": {
            "context_keywords": [
              "BPMN",
              "RPA",
              "Zapier",
              "Make",
              "Power Automate",
              "n8n",
              "UiPath",
              "approval routing",
              "orchestration",
              "triggers",
              "webhooks",
              "ETL",
              "task automation",
              "process mapping",
              "SLA"
            ]
          },
          "maturity": {
            "confidence": 0.86,
            "maturity": "well_known",
            "reasoning": "Workflow automation appears broadly in JDs across ops, IT, and SaaS roles; vendors like Zapier, Power Automate, and ServiceNow show sustained market demand for automating approvals and business processes."
          },
          "skill_id": "workflow-automation",
          "vendor_license": {
            "confidence": 0.99,
            "license": null,
            "vendor": null,
            "year_introduced": null
          },
          "versioning": {
            "current_version": null,
            "version_aliases": {},
            "versioned": false
          }
        },
        "keep_log": [
          {
            "a_dim_id": "workflow-automation-and-approvals",
            "a_name": "Workflow Automation and Approvals",
            "a_role": "__skill_focal__",
            "b_dim_id": "workflow-automation-and-approvals",
            "b_name": "Workflow Automation and Approvals",
            "b_role": "ServiceNOW Developer",
            "pair_kind": "cross_role",
            "reasoning": "Dim A is a general workflow-automation skill cluster with concrete examples like ServiceNow workflows, Power Automate flows, approval chains, task routing, and lifecycle transitions, plus explicit exclusions for scripting and backend orchestration. Dim B is only a ServiceNow Developer role bucket using the same label and a brief role note; it adds no distinct skill content or exemplars. The similarity is from reused wording, not a separate conceptual cluster.",
            "similarity": 0.8113386372207487
          }
        ],
        "locked_dimensions": [
          {
            "description": "Designing and configuring workflow-driven process automation, including approvals, task routing, and lifecycle transitions. Workflow Automation fits here because it is about orchestrating repeatable business or operational steps across people and systems.",
            "exemplar_skills": [
              "Workflow Automation",
              "Approval workflows",
              "Task routing",
              "Process orchestration",
              "Lifecycle automation",
              "ServiceNow workflow design",
              "Power Automate"
            ],
            "in_scope": "Workflow Automation, approval chains, task routing, lifecycle transitions, rule-based process automation, state changes, escalation paths, ServiceNow workflows, Power Automate flows, orchestration of human-in-the-loop steps",
            "name": "Workflow Automation and Approvals",
            "out_of_scope": "Low-level shell scripting, ad hoc cron jobs, and CLI batch tasks; those belong in automation-scripting-and-cli or automation-and-scripting-for-operations. API contract design and backend service orchestration belong in cloud-service-integration-patterns.",
            "overlap_flags": [
              {
                "reason": "Workflow automation often calls external services or APIs, but the core skill is process orchestration rather than integration design.",
                "with_dim_id": "cloud-service-integration-patterns",
                "with_dim_name": null,
                "with_role": "Cloud Architect"
              },
              {
                "reason": "Some workflow automation is implemented with scripts or command-line tools, but this dimension focuses on managed workflow logic and approvals.",
                "with_dim_id": "automation-scripting-and-cli",
                "with_dim_name": null,
                "with_role": "Azure Cloud Engineer, Cloud Engineer"
              }
            ],
            "tentative_id": "workflow-automation-and-approvals"
          },
          {
            "description": "Designing and configuring workflow-driven process automation, including approvals, task routing, and lifecycle transitions. This cluster is coherent because ServiceNow developers often implement process logic rather than standalone application code.",
            "exemplar_skills": [
              "Workflow Automation and Approvals"
            ],
            "in_scope": "Skills, tools, and practices that belong under Workflow Automation and Approvals for the target role, including items implied by the dimension rationale.",
            "name": "Workflow Automation and Approvals",
            "out_of_scope": "Adjacent clusters explicitly not owned by Workflow Automation and Approvals, including unrelated platforms, roles, and skill families per library policy.",
            "overlap_flags": [],
            "tentative_id": "workflow-automation-and-approvals"
          }
        ],
        "merge_log": [],
        "placed": {
          "name": "Workflow Automation",
          "placement_confidence": 0.92,
          "primary_dimension": "workflow-automation-and-approvals",
          "reasoning": "Deterministic JD placement: locked_dimensions has 2 dimension(s) from skill-driven dimension generation after reconciliation; primary_dimension is the first locked dim.",
          "secondary_dimensions": [],
          "skill_id": "workflow-automation"
        },
        "relationships": {
          "child_skills": [],
          "parent_skills": [],
          "related_to": [
            "context-management",
            "migration-scripts",
            "ansible",
            "devops",
            "ci-cd",
            "runbooks",
            "mlops",
            "agentic-systems",
            "powershell",
            "powercli"
          ],
          "requires": [],
          "skill_id": "workflow-automation",
          "suppress_on_match": []
        },
        "skill_id": "workflow-automation",
        "split_log": [],
        "typed": {
          "alternatives_considered": [
            "Tool: ruled out \u2014 the skill name is generic and does not refer to a specific user-operated product.",
            "Concept: ruled out \u2014 it describes an operational process more than a standalone knowledge unit."
          ],
          "confidence": 0.78,
          "name": "Workflow Automation",
          "reasoning": "By the Concept vs Methodology rule, Workflow Automation is best treated as a way of working/process for automating tasks rather than a software product or system shape.",
          "skill_id": "workflow-automation",
          "subtype": "workflow_automation",
          "type": "Methodology"
        },
        "warnings": [
          "stage3_post_filter_dropped_catalog_only_locked_dims:41-\u003e2"
        ]
      },
      "source_tag": "llm",
      "was_in_llm_skills": true
    }
  ],
  "unmatched_skills": [
    "Machine Learning",
    "Document Intelligence",
    "Retrieval",
    "Agentic Workflows",
    "Workflow Automation"
  ]
}

{
  "chosen_role": {
    "display_name": "ServiceNOW Developer",
    "id": 24,
    "rationale": null,
    "role_archetype": null,
    "slug": "servicenow-developer",
    "source": "db"
  },
  "chosen_role_resolution": "in_db",
  "final_input_skills": [
    {
      "skill": "AI",
      "tag": "in_db"
    },
    {
      "skill": "Machine Learning",
      "tag": "new"
    },
    {
      "skill": "Document Intelligence",
      "tag": "new"
    },
    {
      "skill": "Retrieval",
      "tag": "new"
    },
    {
      "skill": "Agentic Workflows",
      "tag": "new"
    },
    {
      "skill": "Workflow Automation",
      "tag": "new"
    }
  ],
  "persistence": {
    "items": [
      {
        "chosen_role_id": 24,
        "dimension": {
          "difficulty_hint": "well_known",
          "display_name": "Version Control Systems",
          "id": 365,
          "rationale": "Tools and workflows for tracking source changes, branching, merging, and collaborating on code history. Git belongs here because it is the canonical distributed version control system used to manage revisions and coordinate team development.",
          "slug": "d_init_01",
          "source": "db"
        },
        "dimension_id": 365,
        "input_skill": "AI",
        "llm_role": null,
        "matched_chosen_role": false,
        "outcome_line": "Existing dimension (library) \u00b7 Role\u2194dimension skipped (dimension not under chosen role)",
        "role_dimension_saved": false,
        "roles_from_db": [],
        "skill_dimension_saved": true,
        "skill_id": 2634,
        "skill_tag": "in_db",
        "skipped_reason": null
      },
      {
        "chosen_role_id": 24,
        "dimension": {
          "difficulty_hint": "well_known",
          "display_name": "Version Control Systems",
          "id": 365,
          "rationale": "Tools and workflows for tracking source changes, branching, merging, and collaborating on code history. Git belongs here because it is the canonical distributed version control system used to manage revisions and coordinate team development.",
          "slug": "d_init_01",
          "source": "db"
        },
        "dimension_id": 365,
        "input_skill": "Machine Learning",
        "llm_role": null,
        "matched_chosen_role": false,
        "outcome_line": "New skill saved \u00b7 Existing dimension (library) \u00b7 Role\u2194dimension skipped (dimension not under chosen role)",
        "role_dimension_saved": false,
        "roles_from_db": [],
        "skill_dimension_saved": true,
        "skill_id": 2672,
        "skill_tag": "in_db",
        "skipped_reason": null
      },
      {
        "chosen_role_id": 24,
        "dimension": {
          "difficulty_hint": "well_known",
          "display_name": "Project Delivery and Coordination",
          "id": 366,
          "rationale": "Coordination practices for organizing work, tracking progress, and aligning stakeholders across a delivery effort. Agile fits here when used as a team execution framework for managing scope, cadence, and collaboration.",
          "slug": "d_init_02",
          "source": "db"
        },
        "dimension_id": 366,
        "input_skill": "Machine Learning",
        "llm_role": null,
        "matched_chosen_role": false,
        "outcome_line": "New skill saved \u00b7 Existing dimension (library) \u00b7 Role\u2194dimension skipped (dimension not under chosen role)",
        "role_dimension_saved": false,
        "roles_from_db": [],
        "skill_dimension_saved": true,
        "skill_id": 2672,
        "skill_tag": "in_db",
        "skipped_reason": null
      },
      {
        "chosen_role_id": 24,
        "dimension": {
          "difficulty_hint": "well_known",
          "display_name": "Version Control Systems",
          "id": 365,
          "rationale": "Tools and workflows for tracking source changes, branching, merging, and collaborating on code history. Git belongs here because it is the canonical distributed version control system used to manage revisions and coordinate team development.",
          "slug": "d_init_01",
          "source": "db"
        },
        "dimension_id": 365,
        "input_skill": "Document Intelligence",
        "llm_role": null,
        "matched_chosen_role": false,
        "outcome_line": "New skill saved \u00b7 Existing dimension (library) \u00b7 Role\u2194dimension skipped (dimension not under chosen role)",
        "role_dimension_saved": false,
        "roles_from_db": [],
        "skill_dimension_saved": true,
        "skill_id": 2673,
        "skill_tag": "in_db",
        "skipped_reason": null
      },
      {
        "chosen_role_id": 24,
        "dimension": {
          "difficulty_hint": null,
          "display_name": "Context Management and Retrieval for LLM Calls",
          "id": null,
          "rationale": "Preparing, selecting, summarizing, and packaging information for model calls so responses stay relevant and grounded. Includes retrieving documents, passages, or structured records; chunking and ranking results; assembling prompt context; and using retrieval-augmented generation, vector search, hybrid search, reranking, and citation grounding to decide what information is included at call time.",
          "slug": "d_merge_01",
          "source": "llm"
        },
        "dimension_id": 264,
        "input_skill": "Retrieval",
        "llm_role": null,
        "matched_chosen_role": false,
        "outcome_line": "New skill saved \u00b7 Existing dimension (reconciliation merge) \u00b7 Role\u2194dimension skipped (dimension not under chosen role)",
        "role_dimension_saved": false,
        "roles_from_db": [],
        "skill_dimension_saved": true,
        "skill_id": 2674,
        "skill_tag": "in_db",
        "skipped_reason": null
      },
      {
        "chosen_role_id": 24,
        "dimension": {
          "difficulty_hint": null,
          "display_name": "Context Management and Retrieval for AI Systems",
          "id": null,
          "rationale": "Preparing, selecting, summarizing, packaging, and retrieving the right context for AI model or agent calls so outputs stay relevant, grounded, and accurate across steps. This includes context-window management, retrieval-augmented generation, memory selection, grounding data, conversation state, tool-result packaging, prompt context assembly, and other mechanisms for passing the right information into model interactions or agentic workflows.",
          "slug": "d_merge_01",
          "source": "llm"
        },
        "dimension_id": 264,
        "input_skill": "Agentic Workflows",
        "llm_role": null,
        "matched_chosen_role": false,
        "outcome_line": "New skill saved \u00b7 Existing dimension (reconciliation merge) \u00b7 Role\u2194dimension skipped (dimension not under chosen role)",
        "role_dimension_saved": false,
        "roles_from_db": [],
        "skill_dimension_saved": true,
        "skill_id": 2675,
        "skill_tag": "in_db",
        "skipped_reason": null
      },
      {
        "chosen_role_id": 24,
        "dimension": {
          "difficulty_hint": "well_known",
          "display_name": "Workflow Automation and Approvals",
          "id": 211,
          "rationale": "Designing and configuring workflow-driven process automation, including approvals, task routing, and lifecycle transitions. This cluster is coherent because ServiceNow developers often implement process logic rather than standalone application code.",
          "slug": "workflow-automation-and-approvals",
          "source": "db"
        },
        "dimension_id": 211,
        "input_skill": "Agentic Workflows",
        "llm_role": null,
        "matched_chosen_role": true,
        "outcome_line": "New skill saved \u00b7 Existing dimension (library) \u00b7 Role\u2194dimension saved",
        "role_dimension_saved": true,
        "roles_from_db": [
          {
            "display_name": "ServiceNOW Developer",
            "id": 24,
            "rationale": null,
            "role_archetype": null,
            "slug": "servicenow-developer",
            "source": "db"
          }
        ],
        "skill_dimension_saved": true,
        "skill_id": 2675,
        "skill_tag": "in_db",
        "skipped_reason": null
      },
      {
        "chosen_role_id": 24,
        "dimension": {
          "difficulty_hint": "well_known",
          "display_name": "Workflow Automation and Approvals",
          "id": 211,
          "rationale": "Designing and configuring workflow-driven process automation, including approvals, task routing, and lifecycle transitions. This cluster is coherent because ServiceNow developers often implement process logic rather than standalone application code.",
          "slug": "workflow-automation-and-approvals",
          "source": "db"
        },
        "dimension_id": 211,
        "input_skill": "Workflow Automation",
        "llm_role": null,
        "matched_chosen_role": true,
        "outcome_line": "New skill saved \u00b7 Existing dimension (library) \u00b7 Role\u2194dimension saved",
        "role_dimension_saved": true,
        "roles_from_db": [
          {
            "display_name": "ServiceNOW Developer",
            "id": 24,
            "rationale": null,
            "role_archetype": null,
            "slug": "servicenow-developer",
            "source": "db"
          }
        ],
        "skill_dimension_saved": true,
        "skill_id": 2676,
        "skill_tag": "in_db",
        "skipped_reason": null
      }
    ],
    "new_skills_created": 5,
    "role_dimension_saved": 0,
    "skill_dimension_saved": 7,
    "skipped": 0
  },
  "planner_output": null,
  "run_id": "cb714ab7-756c-437d-ac85-e5b6557aefa5"
}