# Field Infrastructure Joins > Build field-level offshore infrastructure data bundles from local BSEE tables for engineering products. Use when users ask for field structures, platforms, pipelines, risers, jumpers, umbilicals, FMP/MCP systems, appurtenances, or product-ready field infrastructure joins. - Skill: `vamseeachanta/field-infrastructure-joins` (Agent Skill, multi-file: 2 files) - Install (CLI): `npx skillmds@latest add vamseeachanta/field-infrastructure-joins` - Raw SKILL.md: https://api.skillmd.com/api/skills/vamseeachanta/field-infrastructure-joins/raw - Safety review: pending - Works with: Claude Code, Claude.ai, OpenAI Codex - Category: AI & ML - Author: vamseeachanta (https://skillmd.com/u/vamseeachanta) - Updated: 2026-09-17 - Page: https://skillmd.com/skills/vamseeachanta/field-infrastructure-joins --- # Field Infrastructure Joins Use this skill to turn BSEE field, platform, pipeline, FMP, MCP, scanned-document, and decom tables into product-ready engineering data. ## Quick Start From the `worldenergydata` repo root: ```bash worldenergydata bsee infrastructure-bundle \ --field "Stones" \ --output reports/bsee/field_infrastructure/stones ``` The command defaults to downloaded local data at: `/mnt/ace/worldenergydata/data/modules/bsee/bin` Override the data root when needed: ```bash worldenergydata bsee infrastructure-bundle \ --field "Julia" \ --data-root /mnt/ace/worldenergydata/data/modules/bsee/bin \ --output reports/bsee/field_infrastructure/julia ``` Generate the engineering-facing product package from a bundle: ```bash worldenergydata bsee field-package \ --bundle reports/bsee/field_infrastructure/stones \ --output reports/bsee/field_packages/stones ``` Retrieve source PDFs from a package queue: ```bash worldenergydata bsee retrieve-documents \ --queue reports/bsee/field_packages/stones/document_queue.csv \ --output reports/bsee/field_packages/stones ``` Index retrieved source PDFs for engineering terms: ```bash worldenergydata bsee index-documents \ --manifest reports/bsee/field_packages/stones/download_manifest.csv \ --package-dir reports/bsee/field_packages/stones \ --max-pages 2 ``` OCR no-text source PDFs when scanned drawings need searchable text: ```bash worldenergydata bsee ocr-documents \ --index reports/bsee/field_packages/stones/source_document_index.csv \ --package-dir reports/bsee/field_packages/stones \ --family pipeline_map \ --max-pages 1 ``` Import a filled reviewer-ready template after human source-page review: ```bash worldenergydata bsee import-reviewer-inputs \ --ready-input reports/bsee/field_packages/field_structure_reviewer_ready13_input_template_julia_stones_saint_malo.csv \ --decision-log reports/bsee/field_packages/field_structure_basis_review_decision_log_top15_julia_stones_saint_malo.csv \ --output reports/bsee/field_packages/reviewer_import \ --sqlite-product reports/bsee/field_packages/field_structure_product_julia_stones_saint_malo.sqlite ``` ## Output Contract The command writes: - `field_context.json` - `structures.csv` - `pipeline_segments.csv` - `pipeline_locations.csv` - `appurtenances.csv` - `documents.csv` - `engineering_summary.json` The package command writes: - `index.html` - `document_queue.csv` with `retrieval_url`, `retrieval_status`, and `query_url` columns for BSEE scanned-document review. - `download_manifest.csv` and `source_documents/` when `retrieve-documents` is run against the queue. - `source_document_index.csv` when `index-documents` is run against the retrieval manifest. - `source_document_ocr_index*.csv` when `ocr-documents` is run against no-text indexed PDFs. - Derived `field_document_evidence_*.csv` tables when parser and OCR term-hit rows are consolidated for downstream engineering joins. - Derived `field_engineering_join_input_*.csv` and `field_structure_candidates_*.csv` tables when document evidence is joined back to segment, lease, ROW, appurtenance, and structure context. - Derived `field_structure_review_queue_*.csv` tables when candidates are ranked for manual source-PDF/drawing review. - Derived `field_structure_review_packet_*.html` reports when the top review queue needs local PDF links and product-facing summaries. - Derived `field_structure_review_workbook_*.csv` and `field_structure_review_evidence_*.csv` files when engineering reviewers need decision columns plus per-document evidence snippets. - Derived `field_structure_review_*_geometry*.geojson/.csv/.html` files when the review queue needs spatial screening from pipeline route and appurtenance coordinates. - Derived `field_structure_top*_source_contact_sheet/` packets when reviewers need fast visual triage of the source PDFs behind the highest-priority candidates. - Derived `field_structure_top*_targeted_pages/` packets when reviewers need pages scored by candidate segment IDs and structure terms rather than page-1 thumbnails only. - Derived `field_structure_asset_register_top*_draft_*.csv/.html` files when top candidates need to be exposed as draft register rows with source-page, geometry, and reviewer-decision traceability. - Derived `field_structure_product_manifest_*.json/.html` files when product teams need a stable file catalog, join-key contract, validation status, and promotion gate for a field-structure package. - Derived `field_structure_product_*.sqlite` packages when downstream tools need queryable views across the register, evidence, targeted pages, geometry, segments, appurtenances, and source documents. - Derived `field_structure_engineering_review_batch_*.csv/.html` files when a reviewer needs a one-row-per-candidate work surface generated from SQLite product views. - Derived `field_structure_basis_page_queue_*.csv/.html` files when reviewers need candidate-specific page evidence, thumbnails, excerpts, and blank basis decision fields before promoting register rows. - Derived `field_structure_recommended_basis_page_*.csv/.html` files when reviewers need the top-ranked page per candidate as an acceptance/rejection starting point, without treating the recommendation as verified evidence. - Derived `field_structure_field_join_readiness_*.csv/.html` files when product teams need one row per candidate with field/segment join key, segment context, appurtenance context, geometry coverage, source-document coverage, recommended basis page, and blank field-join review fields. - Derived `field_structure_verified_register_delta_template_*.csv/.html` files when reviewers need a promotion queue with explicit basis acceptance, verified class/quantity/segment fields, reviewer metadata, and follow-up columns before creating a verified register. - Derived `field_structure_basis_acceptance_workpack_*.csv/.html` files when complete recommendation-input rows need to be isolated for immediate source-page acceptance review. - Derived `field_structure_blocked_basis_resolution_queue_*.csv/.html` files when fallback or incomplete basis rows need to be worked separately before promotion. - Derived `field_structure_blocked_basis_resolution_pages/` packets when blocked fallback PDFs need rendered page images, OCR text, contact sheets, and page rankings for manual resolution. - Derived `field_structure_basis_review_decision_log_*.csv/.html` files when product or engineering users need one reviewer-fillable basis decision log across ready and blocked routes. - Derived `field_structure_reviewer_fill_packet_*.csv/.html` files when engineering users need a focused human-fill surface for the minimum fields required by the promotion gate. - Derived `field_structure_reviewer_ready*_input_template_*.csv/.html` files when ready basis-review rows need a spreadsheet-oriented fill template with candidate prefill hints and blank reviewer fields. - Derived `field_structure_promotion_gate_validation_matrix_*.csv/.html` files when products need one row per candidate per required gate field for missing-field audits. - Derived `field_structure_reviewer_input_import_contract_*.csv/.html` files when filled reviewer templates need a source-to-target mapping into the basis decision log. - Derived `field_structure_reviewer_*_staging_audit_*.csv/.html` files when filled reviewer templates need import-readiness checks before updating the decision log. - `reviewer_ready_input_staging_audit.csv`, `basis_review_decision_log_updated.csv`, `promotion_gate_audit.csv`, and `verified_field_structure_register.csv`, plus `import_run_manifest.json` and `index.html`, under the chosen import output directory when `import-reviewer-inputs` validates a filled reviewer template. - `latest_reviewer_import_*` tables and matching `v_latest_reviewer_import_*` views in the SQLite product when `import-reviewer-inputs` is run with `--sqlite-product`. - Derived `field_structure_promotion_gate_audit_*.csv/.html` files when products need fail-closed eligibility checks from pending basis decisions to verified-register promotion. - Derived `field_structure_verified_register_*.csv/.html` files when downstream tools need a stable verified-asset surface that stays empty until human review passes the promotion gate. For column meanings and product usage, read [`references/product-contract.md`](references/product-contract.md). ## Join Order 1. Resolve field leases and area/block anchors from `deepqual/mv_deep_water_field_leases.bin`. 2. Join direct structures from `platstruc`, `permstruc`, FMP, and MCP. 3. Join pipeline-map docs and decom pipeline rows by lease or area/block. 4. Use collected segment IDs to pull route/appurtenance rows from `pipeloc`. 5. Join scanned ROW docs by segment and scanned plan docs by lease. 6. Preserve `source_table`, `join_key`, and `evidence_confidence` in outputs. ## Engineering Use Use this bundle for screening and early engineering workflows: - tieback and host-infrastructure discovery - riser/appurtenance inventory from `PPL_APURT_TYPE` - flowline and umbilical proxy discovery from decom endpoints/product codes - FMP/MCP production measurement context - decommissioning scope discovery - document retrieval queues for maps, plans, and ROW files - source-document term triage for flowline, jumper, riser, umbilical, manifold, PLET/FLET, platform, host, well, and subsea references - OCR triage for image-only pipeline maps and scanned drawings - field-level evidence filtering with boolean term flags such as `has_jumper`, `has_umbilical`, `has_riser`, `has_plet`, and `has_flowline` - segment-level candidate screening with confidence tiers such as `document_text_plus_appurtenance`, `document_text_plus_segment`, and `document_text_nonsegment` - prioritized source-PDF review queues for subsea architecture confirmation - browser review packets for opening BSEE source PDFs and confirming candidate structures - reviewer workbooks with explicit status, verified class, quantity, notes, and follow-up fields - spatial screening layers for route/appurtenance context before source drawing confirmation - source-PDF contact sheets for visual pre-review before filling the verified asset register - targeted page-hit contact sheets to prioritize the source pages most likely to contain candidate structure evidence - draft asset-register rows for product and engineering sessions, with proposed structure class, join keys, source links, targeted pages, geometry references, and blank verification fields - product manifests that let downstream tools ingest the register, evidence, targeted pages, and geometry outputs without rediscovering file paths or join keys - SQLite product views such as `v_asset_product_readiness`, `v_engineering_review_queue`, `v_asset_segment_context`, `v_asset_appurtenance_summary`, `v_asset_source_documents`, and `v_asset_geometry_feature_summary` - SQLite-backed review batches that aggregate source documents, targeted pages, segment context, appurtenance summaries, geometry coverage, review focus, risk notes, and blank decision fields - basis page queues that turn targeted page hits into source-page acceptance worklists for verified document/page decisions - recommended basis page views and exports that reduce the first-pass review set to one page per candidate while preserving the human acceptance gate - field-join readiness exports and `v_field_join_readiness` for products that need a one-row-per-candidate integration surface before verified-register promotion - verified-register delta templates and `v_verified_register_delta_template` for reviewer-fillable promotion queues that keep blocked basis rows separate from rows ready for engineering verification - basis acceptance workpacks and blocker queues that split immediately actionable source-page acceptance from rows needing OCR, query-page, or replacement-basis follow-up - blocked basis resolution page packets that render and OCR all candidate pages for fallback documents before choosing replacement or accepted basis pages - basis review decision logs that unify ready basis-acceptance rows and blocked-resolution rows into a single pending decision surface - reviewer-fill packets that route pending basis decisions to humans with candidate prefill hints while leaving actual acceptance and promotion fields blank - ready-row input templates that isolate immediately actionable basis-review rows and keep reviewer fields blank until source evidence is checked - promotion gate validation matrices that count missing required fields by candidate, route, and field name before rerunning the gate - reviewer input import contracts that map validated fill-template columns back to `basis_review_decision_log` - reviewer staging audits that block decision-log imports until required reviewer fields are present and valid - executable reviewer input imports that update only `import_ready=1` rows and regenerate the fail-closed promotion gate plus verified register outputs after match-only identity fields agree with the target decision log - latest reviewer-import SQLite views that let downstream products query the current import run, staging audit, updated decision log, promotion gate, and verified-register output - promotion gate audits that explain why candidate rows are blocked or ready for verified-register promotion - verified field-structure registers that expose only accepted, reviewer-signed structures and intentionally return zero rows until review is complete Do not claim jumpers/risers/umbilicals are fully enumerated unless the evidence is present in `appurtenances.csv`, `pipeline_segments.csv`, or the scanned document queue/index.