Capability | What the skill does |
|---|
Summarize visible governance inventory | Answers: “In my OAC instance, can you summarize visible governance inventory?” The skill scans visible OAC artifacts returned by the current token, including workbooks, datasets, AI agents, folders, dataflows, connections, models, scripts, and sequences when exposed by the APIs. It summarizes counts, owners, folders, missing metadata, duplicate names, My Folders usage, and other governance review signals. |
Count workbooks per folder | Answers: “How many workbooks are there per folder?” The skill queries workbook/project catalog rows, derives each workbook’s parent folder, counts workbooks by folder, and returns the highest-volume folders first. This helps identify content concentration, heavily used shared folders, and areas that may need ownership or lifecycle review. |
Count distinct workbook datasource references | Answers: “How many distinct data source references are used by workbooks?” The skill lists visible workbooks, fetches workbook details only when needed, extracts datasource references from workbook JSON/XML representations, and reports distinct datasource references, top reused sources, successful detail fetches, and any failures. |
Identify workbooks using scatterplots | Answers: “Which workbooks use scatterplots?” The skill fetches workbook details and searches for visualization tokens such as scatter, scatterplot, and oracle.bi.tech.chart.scatter. It reports confirmed workbook matches, owners, paths, and any detail-fetch failures so the result is evidence-based. |
Find objects using a specific column | Answers: “What objects use the column casualties?” The skill performs a focused column-usage lookup. It searches catalog metadata for the requested field and likely variants, narrows candidate datasets/workbooks/agents/dataflows, inspects only those candidates, and reports exact field-expression matches with object type, owner, path, and column expression. |
Review access granted to a user | Answers: “What objects have I given access to Lalitha?” The skill resolves the named user, identifies the current OAC user, lists objects owned by that user, and inspects access metadata such as roles and ACLs. It separates direct user grants from indirect role-based or inherited access when the API exposes enough evidence. |
Group owned OAC objects by type | Answers: “What OAC objects do I own, grouped by type?” The skill identifies the current OAC user and groups visible owned artifacts by type, such as workbooks, datasets, dataflows, AI agents, folders, connections, models, scripts, and sequences. It can return both summary counts and a detailed object list. |
Count datasources used by my workbooks | Answers: “How many distinct data sources my workbooks relate to?” The skill scopes workbook analysis to workbooks owned by the current user, fetches only those workbook details, extracts datasource references, and reports distinct datasource counts plus workbook-to-datasource mappings. |
Analyze visible dataflows by topic | Answers: “How many visible dataflows are there, and which ones deal with projects data?” The skill uses the purpose-built OAC dataflows endpoint, counts visible dataflows, and searches names, definitions, input datasets, output datasets, and step metadata for project-related terms or other requested topics. |
Find dataflows using Train Model nodes | Answers: “How many dataflows are using train model node?” The skill inspects visible dataflow definitions and applies a strict step test: definition.steps[].type == "TrainModel". It reports both the number of dataflows and the number of Train Model nodes, including owners, names, and model training subtypes when available. |
Generate detailed Excel governance reports | Answers: “Give me a detailed Governance report of all artifacts and objects that I own in emea-field instance in a spreadsheet with different artifacts in separate tabs.” The skill creates an Excel workbook with a Summary tab, Notes tab, and separate tabs for artifact types such as Workbooks, Data Flows, Datasets, Connections, Agents, Folders, Sequences, Scripts, and Models. Reports include owner filtering, artifact paths, datasource references, timestamps, and governance notes. |
Work across OAC instances | Reads oac_url and tokens_path from configuration files such as config/oac-config.json or instance-specific configs like config/emea-field.json, so the same skill can operate across OAC environments without hardcoded tenant URLs. |
Stay read-only by default | Uses read-only OAC APIs for inventory, access review, workbook inspection, dataset metadata, dataflow analysis, and Excel report generation. It does not delete, move, reindex, change permissions, or update artifacts unless the user explicitly requests and confirms a separate remediation action. |
