oracle-ai-data-platform-fusion-bundle

Read-only health + freshness report for the Fusion->AIDP pipeline. Answers 'did the last seed/incremental succeed?', 'what's stale or failed?', 'which marts are materialized?', 'what needs re-running?'. Cross-references fusion_bundle_state (recorded run history) with the LIVE AIDP catalog (what's actually materialized) so a 'success' row next to a missing table reads STALE, and a table with no run record reads UNTRACKED — the live table is the evidence, the state row the overlay. Use when the user asks about pipeline state/health/freshness/last-run, or before deciding what to (re)seed. Does NOT run or change anything (read-only); for running deltas use /aidp-fusion-incremental, for seeding /aidp-fusion-seed.

End-to-end conductor for the Fusion to AIDP to OAC dashboard journey. Takes one high-level goal ('I want a supplier-spend vs GL-balance dashboard from Fusion') and drives the whole chain: configure, connect OAC MCP, bootstrap, seed, advise dataset, author mart if needed, create OAC dataset, author workbook, and optionally enable end-user MCP chat by detecting current state and delegating each step to the right sibling skill/command. Auto-advances on clean steps; pauses for real decisions (destructive seed, variation freeze, OAC dataset creation, ambiguous intent, gaps). Use when the user states a dashboard/analytics goal and wants it driven start-to-finish, OR on first run after installing the plugin: 'I just installed this, what now', 'get me started', 'set me up', 'set up Fusion analytics from scratch / end to end', 'autopilot this', 'take me from nothing to a dashboard'. This is the front door for a fresh install (setup scaffolds the bundle). NOT for a single known step (call that skill directly).

Guide and run the Oracle AIDP Fusion bundle bootstrap phase. Validates bundle/config readiness, checks AIDP/Fusion/IAM prerequisites, runs `aidp-fusion-bundle bootstrap --check-iam` or `bootstrap --refresh`, surfaces tenant variation choices pinned into the active profile YAML and evidence snapshots, and routes unresolved variation failures (`AIDPF-2010` / `AIDPF-2011`) to `/medallion-author`. Use when the user says 'bootstrap', 'pin tenant variation', 'create the profile', 'run pre-seed checks', 'missing profile', 'bootstrap --refresh', or needs the setup phase before seed/incremental. NOT for seed runs, incremental refresh, OAC dataset/workbook work, or net-new mart authoring.

Productized Fusion → Oracle AI Data Platform pipeline with curated BICC extracts (GL/AR/AP/PO/Suppliers/Items), bronze/silver/gold medallion in Delta, conformed dimensions (account/calendar/org/supplier/item), gold marts (AR-Aging/AP-Aging/GL-Balance/PO-Backlog/Supplier-Spend), and MCP-native OAC workbook authoring. Use when the user wants to load Fusion ERP/HCM/SCM data into AIDP, build a CFO dashboard from Fusion, set up a Fusion-backed lakehouse, create OAC datasets/workbooks over AIDP gold, set up OAC MCP for operator authoring or natural-language Fusion analytics in Claude/Cline/Copilot, run BICC extracts incrementally, productize the Oracle blog "Bring Fusion Data into AIDP Workbench Using BICC", or extract Fusion via the saas-batch REST API. Triggers — "load Fusion into AIDP", "set up Fusion bronze layer", "build CFO dashboard from Fusion", "create OAC workbook from Fusion", "run BICC extract", "Fusion AIDP medallion", "saas-batch Fusion extract".

Generate aidp.config.yaml from human-friendly AIDP names instead of hand-copied OCIDs. Use when the user is setting up the fusion-bundle and doesn't want to dig workspaceKey / clusterKey / aiDataPlatformId out of console URLs — they give a workspace name + cluster name (+ region) and this resolves the keys via the AIDP REST API and writes the env block. Triggers — "set up aidp.config.yaml", "configure the fusion bundle connection", "I don't have the OCIDs", "fill in workspace/cluster keys", "what do I put for workspaceKey/clusterKey", "configure AIDP coords by name".