aries-core

ARIES Core

This is the broad ARIES operating skill for Tauris conversion work. It is system-aware but should stay export-format agnostic where possible: reason about ARIES target tables, semantics, keys, and review behavior first; only then decide whether the output should be Access, SQLite, CSV, JSON, or a dry-run memo.

Use this skill when an agent must understand or review how PHDWin v2 data becomes ARIES-ready data through the Tauris.PhdWin conversion model.

Operating Rules

1. Treat ARIES output as contract-driven, not a generic table dump.
2. Treat PHDWin source extraction as read-only. Never mutate native PHDWin files.
3. Prefer SQLite review databases for agent analysis. Use Clarion / TopSpeed ODBC only for the one-time native PHDWin extraction step.
4. Keep the Clarion / TopSpeed driver serialized: one native extraction job at a time per machine/process.
5. For mutating ARIES workflows, produce a dry-run artifact and review summary before writing.
6. Do not invent ARIES field meanings. Mark unknown behavior as unverified.
7. Preserve template-owned ARIES data unless a human explicitly approves taking ownership of it.
8. Preserve unsupported or unknown rows. Do not drop data just because the current converter cannot interpret it.
9. Use Tauris.PhdWin reference logic as the primary behavioral guide for PHDWin-to-Aries conversion.
10. Use deterministic scripts or table checks for repeatable validation whenever available.

When To Load Other Files

Load these only when the task requires the detail:

• references/module-map.md: ARIES area/module routing.
• ../aries-ac-economic/SKILL.md: detailed AC_ECONOMIC parsing, validation, and line editing.
• ../../../phdwin-v2/mcp-servers/PHDWinv2_MCP/PHDWIN_TO_ARIES_TABLE_MAP.md: PHDWin source table map.
• ../../../phdwin-v2/mcp-servers/PHDWinv2_MCP/PHDWIN_TO_ARIES_PLAYBOOK.md: Cowork workflow for PHDWin-to-Aries review.
• ../../../phdwin-v2/mcp-servers/PHDWinv2_MCP/reference/aries-conv-docs/ARIES_ACCESS_TABLE_CONTRACTS.md: fixed ARIES Access table contracts.
• ../../../phdwin-v2/mcp-servers/PHDWinv2_MCP/reference/aries-conv-docs/ARIES_SCHEMA_MAPPING.md: project/member/filter/sort mapping notes.
• ../../../phdwin-v2/mcp-servers/PHDWinv2_MCP/reference/aries-conv-docs/ARIES_EXPORT_RUNNING_LIST.md: latest export-state and known gaps.
• ../../../phdwin-v2/mcp-servers/PHDWinv2_MCP/reference/aries-conv-docs/ARIES_AC_PROPERTY_RULES.md: strict AC_PROPERTY rules.
• ../../../phdwin-v2/mcp-servers/PHDWinv2_MCP/reference/phdwin-v2/Phdwinout definitions_complete.xls: cleared PHDWin output table/field definitions for source interpretation.

Standard Workflow

1. Identify the ARIES target surface: AC_PROPERTY, AC_ECONOMIC, AC_OWNER, AC_PRODUCT, AC_DAILY, AC_TEST, PROJECT, PROJLIST, SelFilters, SORTFILTERS, AC_SCENARIO, AC_SETUPDATA, ARLOOKUP, sidefiles, groups, or Access export.
2. Identify the source layer: native PHDWin .phz/.phd/.mod, exported SQLite, staged CSV, resolved ARIES SQLite/Postgres, or existing ARIES Access template/export.
3. Separate extraction from conversion: native PHDWin -> SQLite review database -> resolved ARIES tables -> Access/CSV/QA outputs.
4. Determine scope: all leases, one LSE_ID, a project, a filter, a sort, or an ownership/partner subset.
5. Check key source tables before mapping: PHD_MAINLSE, PHD_OWNER, PHD_FORCAST, PHD_MONHIST, PHD_GROUPS, PHD_LIST, PHD_PRODUCTNAMES.
6. Build or review the ARIES target rows against fixed target table contracts.
7. Validate keys, row counts, missing memberships, duplicate target keys, and template-owned/additive table behavior.
8. Return a concise memo with source path, scope, tables reviewed, conversion risks, and recommended next action.

Conversion Architecture

The preferred public workflow is SQLite-first:

PHDWin .phz / .phd + .mod
  -> one serialized Clarion / TopSpeed ODBC extraction
  -> SQLite review database
  -> resolved ARIES table rows
  -> named CSV files and/or Access template copy
  -> human review in ARIES

Do not design agent workflows that repeatedly query native PHDWin through the Clarion driver. Once SQLite exists, all deeper review should use SQLite or staged table files.

The resolved ARIES layer should be scoped and replaceable:

• ConvertAll
• ConvertLease(LSE_ID)
• later ConvertProject(PROJKEY)
• later ConvertFilter(FLT_ID)
• later ConvertSort(SRT_ID)

For ConvertLease(LSE_ID), delete/rebuild only lease-scoped resolved rows for that lease and leave unrelated rows intact.

PHDWin Source Tables

Use these PHDWin tables as the main conversion inputs:

PHDWin source	Conversion role
PHD_TITLES	dataset title/project header/default context
PHD_MAINLSE	core case/well/property source; primary LSE_ID anchor
PHD_PRODUCTNAMES	product code labels
PHD_OWNER	ownership and partner context
PHD_GROUPS	group/project/incremental definitions
PHD_LIST	explicit project/group membership
PHD_ADJOWNER	adjusted ownership review
PHD_FORCAST	forecast rows; preserve source spelling
PHD_LSEPRODVAL	case/product value overrides
PHD_LSESEGMENT	lease segment and forecast/economic detail
PHD_MONHIST	monthly production history
PHD_CUMVOL	cumulative volume context
PHD_ECON	economic assumptions
PHD_INVEST	investment/capital rows
PHD_INVESTDESCR	investment labels/descriptions
MOD_SCEN	model/scenario assumptions when present
MOD_TEMPLATE	model templates and assumptions when present
PHD_FILTER, PHD_FILTERLINE	saved filter definitions
PHD_SORT	saved sort/subtotal definitions
PHD_CLASS, PHD_CATEGORY	reserve class/category lookup
PHD_IDCODES, PHD_IDLABELS	field/operator/basin/custom code lookup context

Missing optional tables should be reported as risk or limitation, not fatal by default. Missing core tables such as PHD_MAINLSE, PHD_OWNER, PHD_FORCAST, or PHD_MONHIST are conversion risks.

ARIES Target Areas

ARIES target	Primary source
AC_PROPERTY	PHD_MAINLSE, title/context, reserve class/category, ID lookup tables
AC_OWNER	PHD_OWNER, PHD_GROUPS, PHD_ADJOWNER
GROUPTEST	group/member/test shape only when source rows match template requirements
AC_PRODUCT	PHD_MONHIST, product names, forecast/history context
AC_DAILY	daily source rows when present; do not duplicate test rows as daily production
AC_TEST	test/daily test source rows when present
AC_ECONOMIC	forecast, segment, product value, econ, investment, model scenario/template logic
PROJECT	default project, partner projects, PHDWin groups/incrementals
PROJLIST	project membership from PHD_LIST, fallback from ownership when necessary
SelFilters	project-level selection/filter behavior
SORTFILTERS	project-level sort stack behavior
AC_SCENARIO	group/model scenario data
AC_SETUPDATA	generated Tauris setup rows appended to template rows
ARLOOKUP	code/product/class/category lookup rows
sidefiles	resolved sidefile/economic auxiliary data

Key Anchors

The source key is usually LSE_ID. Preserve it through diagnostics and resolved rows whenever practical.

Important source joins:

• case/well: PHD_MAINLSE.LSE_ID
• ownership: PHD_OWNER.LSE_ID, PHD_OWNER.GRP_ID, PHD_OWNER.SEQ
• groups/projects: PHD_GROUPS.GRP_ID, PHD_LIST.GRP_ID, PHD_LIST.LSE_ID
• forecasts: PHD_FORCAST.LSE_ID, ARCSEQ, PRODUCTCODE
• history: PHD_MONHIST.LSE_ID, TYPE, YEAR
• filters: PHD_FILTER.FLT_ID, PHD_FILTERLINE.FLT_ID
• sorts: PHD_SORT.SRT_ID

Important ARIES target keys:

Table	Effective key
AC_PROPERTY	DBSKEY, PROPNUM
AC_PRODUCT	PROPNUM, P_DATE
AC_DAILY	PROPNUM, D_DATE
AC_TEST	PROPNUM, T_DATE
AC_ECONOMIC	PROPNUM, SECTION, SEQUENCE
AC_OWNER	PROPNUM, SCENARIO, PHASENAME, STARTDATE
PROJECT	PROJE, PROJKEY
PROJLIST	PROJKEY, PROJSEQ
AC_SCENARIO	DBSKEY, SCEN_NAME, DATA_SECT
AC_SETUPDATA	non-keyed/additive
ARLOOKUP	NAME, LINETYPE, SEQUENCE

Project Model

Tauris.PhdWin conversion logic treats ARIES project behavior as first-class. Do not treat groups, filters, and sorts as decorative metadata.

Use three project types:

1. Default dataset project:
   • project key: 00_RSV_CAT
   • user-facing intent: All Cases
   • synthetic default, not a direct copy of source group All Cases
2. Partner projects:
   • ownership-driven
   • derive from qualified ownership context
   • not the same thing as PHDWin group/incremental objects
3. Incremental/group projects:
   • source: PHD_GROUPS plus PHD_LIST
   • represent economic difference entities
   • may include visible incremental member cases

Priority chain:

PHD_GROUPS / PHD_LIST / PHD_OWNER
  -> PROJECT
  -> PROJLIST
  -> AC_PROPERTY.PROPNUM
  -> SelFilters
  -> SORTFILTERS

Build project membership from PHD_LIST first when it exists. Use PHD_OWNER SEQ = 1 only as fallback membership source. Keep ownership as qualified-interest logic, not the primary membership list.

AC_PROPERTY Rules

AC_PROPERTY is the highest-priority structural table.

Hard rules:

• table name must be AC_PROPERTY
• final Access columns are uppercase
• no spaces or punctuation in final columns
• preserve underscores where needed
• required key pair: DBSKEY, PROPNUM
• treat DBSKEY + PROPNUM as unique
• do not derive final column casing from C# property names or dictionary keys

Core structural fields include:

• DBSKEY
• PROPNUM
• SEQ
• MAJOR
• PRIOR_OIL
• PRIOR_GAS
• PRIOR_WTR

Use the template as schema guidance, but Tauris export rules override loose dynamic naming.

AC_ECONOMIC Rules

Use the narrower aries-ac-economic skill for detailed line parsing and validation.

At this level, remember:

• AC_ECONOMIC is lease-scoped.
• key shape is PROPNUM, SECTION, SEQUENCE.
• forecast rows come primarily from PHD_FORCAST.
• production/economic detail may also require PHD_LSESEGMENT, PHD_LSEPRODVAL, PHD_ECON, PHD_INVEST, PHD_INVESTDESCR, MOD_SCEN, and MOD_TEMPLATE.
• sidefile expansion, lookup expansion, setup common/default lines, scenario selection, and macro substitution are fidelity risks.
• unsupported economic rows should be preserved unless a human explicitly approves dropping or rewriting them.

If asked to generate or edit actual economic lines, load aries-ac-economic and its references before drafting output.

Actuals: AC_PRODUCT, AC_DAILY, AC_TEST

Use production history and actuals tables carefully:

• AC_PRODUCT is primarily monthly product/history output.
• AC_TEST is test data output.
• AC_DAILY is daily production output when a source daily production table exists.
• Do not duplicate test rows into AC_DAILY as production.
• If daily source rows are absent, report that AC_DAILY cannot be populated from current evidence rather than fabricating rows.

For each lease-scoped output table, row counts should be reconcilable to selected LSE_ID scope.

Ownership And Groups

Ownership is not merely project membership.

Use PHD_OWNER for qualified interests, partner context, phases, scenarios, and AC_OWNER output. Use PHD_GROUPS and PHD_LIST for group/incremental project behavior.

Review risks:

• PHD_MAINLSE rows with no ownership rows
• owners/groups that do not join cleanly
• duplicate owner rows for the same target key
• adjusted ownership rows in PHD_ADJOWNER that change the intended output
• named groups that share visible lease membership but differ by ownership or economic context

GROUPTEST is lower priority than project membership and should be skipped unless source rows match the Access table shape.

Filters And Sorts

SelFilters and SORTFILTERS describe project behavior in ARIES. Their meaning depends on PROJECT, PROJLIST, and AC_PROPERTY.

SelFilters commonly includes:

• ProjKey
• SeqNum
• TableAlias
• TableColumn
• Operator
• OperatorText
• AndOr
• DataType

Known filter examples:

• TAI_EXCLUDE is Null
• RSV_CAT is one of PDP, PUD, PROB
• LSE_ID is one of 2.00

SORTFILTERS commonly includes:

• ProjKey
• SeqNum
• TableAlias
• TableColumn
• SortOrder
• SortBreak

Known sort columns:

• CLASS
• RSV_CLASS
• RSV_CAT
• STATE
• FIELD
• LEASE
• RSC_SORT
• LSE_ID

When explicit sort/filter logic is incomplete, generate stable starter rows rather than pretending to infer the last active PHDWin UI state. Default sort stack can start from reserve category/class behavior such as RSV_CAT when no explicit source is available.

Scenario And Setup Data

AC_SETUP is template-owned and should generally not be rebuilt.

AC_SETUPDATA is additive:

• preserve template-defined rows
• append generated Tauris setup rows
• do not truncate the table unless explicitly instructed

AC_SCENARIO should be generated from group/model scenario context when present. Validate DBSKEY, SCEN_NAME, and DATA_SECT uniqueness where practical.

Access Export Behavior

For Aries Access .accdb export:

1. Copy the template database first.
2. Prefer writing into existing template tables when their structure is authoritative.
3. Use fixed Aries table names and explicit column mappings.
4. Do not infer table names from DTOs, filenames, JSON keys, or C# property names.
5. Preserve template-owned tables such as AC_SETUP.
6. Append to additive tables such as AC_SETUPDATA.
7. Recreate or write export-owned lease tables from resolved ARIES rows: AC_PROPERTY, AC_PRODUCT, AC_TEST, AC_DAILY, AC_ECONOMIC.
8. Validate duplicate effective keys before insert where practical.
9. Preserve row order where order is part of the contract, especially PROJLIST.
10. Treat Access ODBC and column/type coercion failures as export issues, not source-data truth.

Resolved ARIES tables should be the preferred export contract when available. The Access writer should not rebuild semantic conversion ad hoc if a resolved layer already exists.

Date Handling

PHDWin dates may be Clarion dates. Use:

date = 1800-12-28 + raw_clarion_days

Example:

82123 -> 11/01/2025

Show converted dates in diagnostics when possible, not only raw values.

Review Queries For SQLite

Use these as starting points when a PHDWin SQLite review database is available.

Table inventory:

SELECT name
FROM sqlite_master
WHERE type = 'table'
ORDER BY name;

Core row counts:

SELECT 'PHD_MAINLSE' AS table_name, COUNT(*) AS row_count FROM PHD_MAINLSE
UNION ALL SELECT 'PHD_OWNER', COUNT(*) FROM PHD_OWNER
UNION ALL SELECT 'PHD_GROUPS', COUNT(*) FROM PHD_GROUPS
UNION ALL SELECT 'PHD_LIST', COUNT(*) FROM PHD_LIST
UNION ALL SELECT 'PHD_FORCAST', COUNT(*) FROM PHD_FORCAST
UNION ALL SELECT 'PHD_MONHIST', COUNT(*) FROM PHD_MONHIST;

Cases missing ownership:

SELECT M.LSE_ID, M.LSE_NAME
FROM PHD_MAINLSE M
LEFT JOIN PHD_OWNER O
  ON O.LSE_ID = M.LSE_ID
WHERE O.LSE_ID IS NULL
ORDER BY M.LSE_NAME;

Group membership:

SELECT L.GRP_ID, COUNT(*) AS member_count
FROM PHD_LIST L
GROUP BY L.GRP_ID
ORDER BY L.GRP_ID;

Forecast product coverage:

SELECT PRODUCTCODE, COUNT(*) AS row_count
FROM PHD_FORCAST
GROUP BY PRODUCTCODE
ORDER BY PRODUCTCODE;

Filter and sort inventory:

SELECT 'PHD_FILTER' AS table_name, COUNT(*) AS row_count FROM PHD_FILTER
UNION ALL SELECT 'PHD_FILTERLINE', COUNT(*) FROM PHD_FILTERLINE
UNION ALL SELECT 'PHD_SORT', COUNT(*) FROM PHD_SORT;

Conversion Readiness Memo

When asked to review a PHDWin-to-Aries conversion, return a memo with:

• source path and source type
• whether review used native PHDWin, SQLite, CSV, or Access
• driver/readability status if native files were involved
• core table presence and row counts
• selected lease/project/filter/sort scope
• project membership interpretation
• AC_PROPERTY readiness
• ownership readiness
• actuals readiness: AC_PRODUCT, AC_DAILY, AC_TEST
• economics readiness: AC_ECONOMIC
• setup/scenario/lookup readiness
• duplicate key or missing join risks
• known limitations and unverified assumptions
• recommended next action

Red Flags

Call these out explicitly:

• native driver required but missing
• concurrent native driver use
• missing PHD_MAINLSE
• missing PHD_OWNER
• missing PHD_FORCAST
• missing PHD_MONHIST
• PHDWin source can be inventoried but cannot be queried
• PHD_LIST absent when group/project fidelity is important
• duplicate target keys in any ARIES table
• AC_PROPERTY column casing drift
• dynamic Access column names replacing fixed contracts
• template-owned rows being deleted
• AC_SETUPDATA being truncated instead of appended
• daily/test rows being misclassified
• group/incremental projects treated as plain folders
• last-used PHDWin UI filter/sort state inferred without evidence

Output Preference

For public Claude Cowork workflows:

1. SQLite review database first.
2. Named CSV files for inspectable ARIES target tables.
3. Optional Access .accdb export from the bundled template when Windows Access ODBC is available.
4. Human review before any ARIES import or mutation.