legacy-spec-factory

Orchestrate IBM i program analysis from a program-list CSV or Excel export for Copilot Chat-limited environments. Use when the user has many RPGLE/SQLRPGLE/CLLE/COBOL programs to scan independently and needs one-program-per-chat prompt queues, per-program status tracking, resumable handoff, or batch manifests. Supplemental skill; delegates actual source analysis to legacy-ibmi-program-analyzer and leaves any later cross-program/program-set review to legacy-ibmi-flow-analyzer after a specific flow or program set is selected.

Merge multiple IBM i program-analysis artifacts into a compact SME program-set core review, especially when an SME provides a program-flow list and wants Calculation Logic, Validation Logic, Exception Handling, and Message Inventory. Default field workflow is program-evidence first with no cross-run reuse: analyze every distinct SME-provided program for the current run, reuse only artifacts produced earlier in the same run/batch, then assemble the self-contained program-set review. Also supports approved document-repo reuse after an all-program scan has been merged: SME can clone the document repo locally, provide a program flow, and assemble a core review from approved existing program artifacts without re-scanning source. Supports source inventory cache reuse from outputs/repo-scan when fresh and explicit full transaction-flow analysis for seven trigger models when requested separately. Layer 1.5 (platform-specific) skill of the Legacy Spec Factory reverse chain.

Analyze individual IBM i programs (RPGLE, CLLE, COBOL) to extract control flow, file I/O, external calls, and error handling with evidence backing. Use when diving deep into one program's behavior from an approved inventory, when standalone exploratory review is needed, or when checking whether an existing central delivery artifact can be reused before scanning source again. Layer 1 (platform-specific) skill of the Legacy Spec Factory reverse chain.

Extract evidence-backed current-state functional discovery packages from selected RAG/document evidence, SME notes, spreadsheets, and project folders. Use when a team needs a business-facing functional discovery report plus structured function, validation, calculation, interface, channel/report, accounting-impact, traceability, and gap catalogs from reviewed documents. Routes raw binary documents to legacy-document-evidence-intake, RAG packaging to legacy-module-context-intake, detailed IBM i code/program-flow analysis to legacy-ibmi-program-analyzer or legacy-ibmi-flow-analyzer, and final handoff consolidation to a downstream output generator.

Inventory IBM i / AS400 legacy assets for modernization. Use when collecting or reviewing RPGLE, CLLE, COBOL, DDS, DB2 for i, jobs, screens, reports, spool, and runtime evidence before generating an evidence-backed spec. Layer 1 (platform-specific) skill of the Legacy Spec Factory reverse chain.

Assemble and validate an IBM i business module from approved flow analyses or a ready module-first context package, producing the evidence-backed module overview plus Program Flow and Data Flow artifacts. Operation Flow and System Flow are no longer default outputs because they require stronger SME or architecture evidence than most code-backed runs provide. Use when multiple flows belong to the same business module, or when `legacy-module-context-intake` has normalized external RAG / human context and you need evidence-bounded module assembly before BRD writing and review. Layer 1.5 (platform-specific) skill. Implements the model defined in `docs/module-analysis-model.md`.

Package external RAG or code-knowledge-graph output, document-intake manifests, source metadata, SME fragments, and human-confirmed module context into a traceable `00_context_packages/<MODULE-SLUG>/` package. Use when a team has module-first context, document evidence intake output, source snippets, field dictionary mappings, impact scope, contradictions, retrieval gaps, reviewed module notes, or sparse source metadata and needs a safe handoff to `legacy-ibmi-module-analyzer` or BRD preparation. Blocks on unauthorized or unredacted evidence, hidden contradictions, or attempts to promote RAG candidates, generated-draft context, or sparse-context TBDs into approved business rules, BRD facts, or canonical module flows; treats rough scope and degraded context as warnings/TBDs.

Export stakeholder-facing Markdown artifacts into standalone HTML companions for easier reading in pilot reviews, SME walkthroughs, and business sign-off sessions. Use when the user asks for HTML versions of BRDs, specs, review packs, question packs, traceability pages, STATUS pages, or other user / SME-facing docs. Supplemental skill only: Markdown remains the canonical source, and this skill generates `.html` companions plus optional directory `index.html` pages without redefining the underlying artifact contract.

Governance and quality-contract skill for every step in the Legacy Spec Factory reverse chain. Use when authoring, executing, or reviewing any pipeline step (inventory, program analysis, flow analysis, module analysis, spec writing, spec review, forward SDLC handoff) and you need a shared INPUT → EXECUTION → OUTPUT → VALIDATION contract. This is not a business analyzer — it does not produce inventory, program analyses, specs, or reviews itself. It defines, checks, and reports on the quality contract those skills must satisfy.

Validate any Legacy Spec Factory step artifact against the INPUT → EXECUTION → OUTPUT → VALIDATION contract. Use after a step has produced its draft artifact (inventory, program analysis, flow, module, spec, or handoff bundle) and you need a structured `pass` / `pass_with_warnings` / `blocked` review report with mechanical, semantic, and SME-readiness findings. This skill does not produce business artifacts itself and does not replace SME approval — it surfaces what is mechanically valid, what needs semantic review, and what only an SME can decide.

Export stakeholder-facing Markdown artifacts into standalone HTML companions for easier reading in pilot reviews, SME walkthroughs, and business sign-off sessions. Use when the user asks for HTML versions of BRDs, specs, review packs, question packs, traceability pages, STATUS pages, or other user / SME-facing docs. Supplemental skill only: Markdown remains the canonical source, and this skill generates `.html` companions plus optional directory `index.html` pages without redefining the underlying artifact contract.

Export stakeholder-facing Markdown artifacts into standalone HTML companions for easier reading in pilot reviews, SME walkthroughs, and business sign-off sessions. Use when the user asks for HTML versions of BRDs, specs, review packs, question packs, traceability pages, STATUS pages, or other user / SME-facing docs. Supplemental skill only: Markdown remains the canonical source, and this skill generates `.html` companions plus optional directory `index.html` pages without redefining the underlying artifact contract.

Validate and package one Atlas-compatible SDD handoff from an approved Legacy BRD and approved Legacy spec (`spec.yaml`/`spec.md`). Bridge / gate / package skill — does not generate code, architecture, design, user stories, or test code. Enforces `docs/forward-sdlc-contract.md`, refuses to bypass `legacy-spec-writer`, and produces machine-readable + human-readable handoff artifacts plus a strict findings report. Layer 2 (platform-agnostic).

Use when migration discovery needs an evidence-backed legacy-system BRD for one capability, a daily-delivery BRD draft with exception-only review, or an internal POC BRD draft from sparse/partial context so stakeholders can review scope early. Produces SME-reviewable observed behaviors, inferred rule seeds, open gaps, traceability, and BRD-stage validation scenario seeds. This is the primary near-term old-system discovery artifact; it does not compare against the new system, create an SDD handoff, or mandate target-system changes.

Extract structured `observed_in_runtime` evidence from approved IBM i job logs and spool/report files into `runtime-evidence.jsonl`. Use after `legacy-ibmi-evidence-intake` has approved the evidence manifest and `legacy-ibmi-inventory` can map runtime artifacts to `OBJ-*` IDs. Blocks on missing approval, unredacted confidential evidence, or missing inventory mappings; never infers business rules or modernization decisions from runtime logs.

Entry-point router for the Legacy Spec Factory reverse chain. Use for natural-language guidance through IBM i / AS400 / RPGLE / CLLE / COBOL modernization, including document/spec/RAG/context intake, inventory, program and flow analysis, module analysis, legacy BRD discovery, post-BRD old-vs-new disposition, spec writing, and SDLC handoff. Routes users to the safest next skill and does not replace downstream extraction, synthesis, review, risk-assessment, or gap-analysis processes.

Package external RAG or code-knowledge-graph output, document-intake manifests, source metadata, SME fragments, and human-confirmed module context into a traceable `00_context_packages/<MODULE-SLUG>/` package. Use when a team has module-first context, document evidence intake output, source snippets, field dictionary mappings, impact scope, contradictions, retrieval gaps, reviewed module notes, or sparse source metadata and needs a safe handoff to `legacy-ibmi-module-analyzer` or BRD preparation. Blocks on unauthorized or unredacted evidence, hidden contradictions, or attempts to promote RAG candidates, generated-draft context, or sparse-context TBDs into approved business rules, BRD facts, or canonical module flows; treats rough scope and degraded context as warnings/TBDs.

Produce evidence-backed `spec.yaml` and `spec.md` artifacts from approved module + flow + program analyses plus an approved legacy-system BRD Package after stakeholders explicitly promote a capability beyond legacy discovery. One spec per business capability (CAP-*). Layer 2 (platform-agnostic) skill — sits after BRD review and any separate post-BRD comparison / gap-analysis decision, producing the contract that downstream SDLC can consume. Implements `schemas/spec.schema.yaml`.

Package, validate, and audit one capability's end-to-end traceability across the Legacy Spec Factory reverse chain — evidence manifest, inventory, program/flow/module analyses, optional BRD, approved `spec.yaml`/`spec.md`/`traceability.md`, and optional SDD handoff. Audit / packager / gate skill — does not mint business rules, behaviors, acceptance criteria, modernization decisions, evidence, test cases, or SDD handoffs, and does not promote review status. Emits a sealed traceability package or a strict blocked-findings report. Layer 2 (platform-agnostic).

Use when a BRD, spec, module analysis, or other Legacy Spec Factory artifact needs documented SME validation, especially when the SME should answer guided review questions directly in chat and have decisions written back to review artifacts.

Intake and format-normalize legacy enterprise documents (Excel .xlsx/.xlsm/.xls, Word .docx/.doc, PowerPoint .pptx/.ppt, Visio .vsdx/.vsd, PDF, images, screenshots, scanned docs) into Markdown, CSV, PDF, PNG/SVG, and manifests with evidence coordinates and quality gates before legacy-module-context-intake. Use when business/technical documents are in formats that GitHub Copilot or downstream LLM skills cannot reliably consume, when legacy binary Office or Visio files need conversion, or when macros, scanned pages, or unauthorized production data require security/redaction review. Does not infer business rules, generate BRD/spec content, approve evidence, or hide contradictions. Routes to legacy-ibmi-evidence-intake when sensitivity or authorization is unresolved; otherwise hands off to legacy-module-context-intake.

Aggregate every per-program analysis under one module into a single SME-friendly batch digest table (`programs-batch-digest.md`). Reduces SME review friction for medium/large modules — a 50-program module turns from 50 files to one scannable page grouped by criticality (critical / standard / low_risk), with one-line roles, key pending decisions, TBD counts, and links to detail. Trigger when the user says "give me the SME-facing program review", "我要给 SME 看程序清单", "batch review the programs in this module", or after a batch of new program-analysis.md files lands. Supplemental skill — does not advance the linear stage_id; co-exists with `legacy-ibmi-program-analyzer` (which produces the detail) and `legacy-sme-review-facilitator` (which builds the active decision queue).

Produce an evidence-backed old-vs-new golden master test plan from an approved Legacy spec, approved acceptance criteria, and redacted runtime evidence. Use when a capability has `spec.yaml.status: approved` and the team needs `TC-*` golden master cases, sample-data references, expected-output references, and equivalence coverage before forward SDLC or implementation validation. Planning skill only: it does not generate executable test code, run comparisons, collect raw production data, or invent expected outputs.

Analyze IBM i data models from DDS physical/logical files, DB2 for i metadata, SQL DDL, approved inventory, and program/flow evidence to produce an evidence-backed data model package with dictionary, access paths, CRUD matrix, and SME review checklist. Use when reverse-engineering a business domain's data structure to support modernization. Layer 1.5 (platform-specific) skill of the Legacy Spec Factory reverse chain.

Register, classify, assign evidence IDs, and record source-path authorization or redaction governance for IBM i modernization evidence. Use when preparing source code, DB metadata, job logs, spool files, screen samples, reports, transactions, or runtime evidence before inventory analysis. This skill records intake metadata, source paths, internal review decisions, and redaction approvals when required. Precedes legacy-ibmi-inventory.

Analyze IBM i DSPF, PRTF, menus, subfiles, function keys, indicators, screen samples, spool/report samples, and presentation-layer behavior with evidence backing. Use when screen/report artifacts need structured analysis before or alongside program/flow analysis. Layer 1 (platform-specific) skill of the Legacy Spec Factory reverse chain.

Produce, expand, review, and validate modernization decision records (DEC-*) when decisions become too large or risky to keep only inside spec-writer. Optional Governance / BRD skill for structured decision packages. Bridges between approved BRD/spec and target-platform authority approval. Does not recreate Atlas SDD architecture/design artifacts.

Execute and validate runtime smoke tests for legacy spec factory skills across Codex, Claude Code, and OpenCode. Use when a new skill is added or an existing skill materially changes, and you need to verify portability and move it from `synced` to `passed` in docs/runtime-matrix.md.