anti-legacy-deliverables

星标1

分支0

更新时间2026年6月19日 05:52

Render the full stakeholder DELIVERABLES package once the requirements graph is ready — product requirements (PRD), architecture diagrams, detailed test strategy, functional test scripts, end-to-end migration plan, and the living risk / decisions / evidence logs — into .anti-legacy/deliverables/, then compile an index. Register-only; does not advance the pipeline. Use when: "produce the deliverables", "generate all the deliverables", "the deliverables package", "PRD + diagrams + plan + risk/decision logs".

安装

用 Codex 或 Claude 帮你安装复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。

来源

mikeparcewski

mikeparcewski/anti-legacy

打开 GitHub 仓库查看创作者相关仓库

下载

相关职业SOC

基于 SOC 职业分类

软件开发工程师计算机与数学类职业·SOC 15-1252

文件资源管理器

2 个文件

SKILL.md

readonly

同仓库更多 Skills

同仓库

anti-legacy-graph-validator

mikeparcewski/anti-legacy

Validates the requirements graph against the wicked-estate code graph before the blueprint phase. Checks every requirement's legacy field resolves to a real program node in the graph (not a JCL step), finds uncaptured behavior-bearing nodes with business logic that need requirements, and flags duplicate program references. Blocks if errors are found; warns and requires human review if only gaps remain. Run after extraction/graph-translator, before blueprint. Use when: "validate requirements graph", "audit the graph", "check graph accuracy", "are we missing any programs", "graph audit before dev", "validate before blueprint".

2026-06-191

anti-legacy-analyze

mikeparcewski/anti-legacy

Apply structural lenses to the legacy code graph to identify entry points, domain boundaries, shared data coupling, operational modes (batch vs online), and risk hotspots. Queries the wicked-estate code graph via the helper (never raw SQLite, never legacy_graph.json). Produces an analysis report. Use when: "analyze the legacy code", "identify entry points", "find domain boundaries", "what are the hotspots", "classify the programs".

2026-06-191

anti-legacy-antagonist

mikeparcewski/anti-legacy

Pre-build threat modeler — the adversarial half of the Phase Execution Protocol (PEP, AGENTS.md §10). Given a phase plan and its threat surface, generates a structured threat list BEFORE the producer runs so the producer can address each threat in its output rather than discovering the gap post-build. CRITICAL threats block the phase from advancing until the producer acknowledges them (fix or documented waiver). MEDIUM/MINOR are advisory. Distinct from anti-legacy:adversarial-review (which reviews rendered output AFTER production) — the antagonist attacks the plan BEFORE production, while it is still cheap to change. Use when: any phase producer is about to run; "antagonist check"; "threat model this phase"; "what could go wrong"; PEP step 3 in orchestrate.

2026-06-191

anti-legacy-expert

mikeparcewski/anti-legacy

The anti-legacy pipeline internals SME, and the home of the shared antilegacy_core Python library. Use when: understanding or operating the pipeline core, diagnosing a core module (estate seam, requirements-graph builder, coverage, manifest state machine, validator), checking pipeline readiness, or extending antilegacy_core.

2026-06-191

anti-legacy-blueprint

mikeparcewski/anti-legacy

Map the Requirements Graph to a target state architecture blueprint. Defines the modern package structure, API surface (REST/gRPC), database schemas, class boundaries, and repository patterns for the target stack. Produces blueprint.json + blueprint.md. Use when: "design the target architecture", "create the blueprint", "map requirements to target stack", "design the Spring Boot structure", "what does the modern version look like".

2026-06-191

anti-legacy-deploy

mikeparcewski/anti-legacy

Package and deploy the modernized application to the configured deployment target. Supports GCP Cloud Run, AWS ECS/Fargate, Azure AKS, and generic Kubernetes. Generates Dockerfile, CI/CD pipeline config, and deployment manifests. Requires GATE_4_UAT to be cleared. Use when: "deploy the app", "generate the Dockerfile", "create CI/CD config", "package for Cloud Run", "generate Kubernetes manifests", "deployment".

2026-06-191

name

anti-legacy:deliverables

description

anti-legacy:deliverables

The umbrella that produces the human-facing deliverables package from the pipeline's structured artifacts. It runs when the graph is ready (after graph-translate, ideally with blueprint + test-strategy available) and writes everything under .anti-legacy/deliverables/.

This is not the review-packet (that is the single offline GATE_1 review document). These are the detailed, individually-consumable deliverables a delivery team needs:

Deliverable	Skill	Output
Product Requirements (PRD)	`anti-legacy:prd`	`deliverables/product-requirements.md`
Architecture Diagrams (Mermaid)	`anti-legacy:diagrams`	`deliverables/diagrams/*.mmd` + index
Test Strategy (data-parity / UAT / E2E / API)	`anti-legacy:test-plan`	`deliverables/test-strategy.md`
Functional Test Scripts	`anti-legacy:test-scripts`	`deliverables/tests/{data-parity,uat,e2e,api}/…`
Migration Plan (epics→subtasks) + Jira CSV	`anti-legacy:migration-plan`	`deliverables/migration-plan.md` + `.jira.csv`
Risk Log (living)	`anti-legacy:risk-log`	`deliverables/risk-log.md`
Decisions Log / ADRs (living)	`anti-legacy:decisions-log`	`deliverables/decisions-log.md`
Evidence Log with receipts (living)	`anti-legacy:evidence-log`	`deliverables/evidence-log.md`

Each deliverable registers its artifact (deliverable-*) in the manifest and never advances the phase — phase advancement stays owned by the phase skills. The living deliverables (risk / decisions / evidence) reflect current state, so re-run them at each gate.

Cross-Platform Notes

Every command runs through the dispatcher (python3 .anti-legacy/run.py <stem>), which is pure Python — identical on macOS, Linux, WSL, and Windows.

When it runs / prerequisites

Required: .anti-legacy/requirements/requirements_graph.json (produced by anti-legacy:graph-translator). Without it the renders refuse to write hollow output.
Enriches when present (degrade gracefully when absent, naming the gap): blueprint.json (diagrams, test-plan, migration-plan), contracts/ (test-plan, test-scripts), the manifest + audit.jsonl (decisions-log, evidence-log), coverage-report.json + annotations.jsonl (risk-log).
Natural home in the pipeline: the review-packet stage (graph + blueprint + test-strategy ready, feeding GATE_1). Also runnable standalone anytime the graph exists.

Step 1: Verify the graph is ready

python3 .anti-legacy/run.py manifest status

Confirm .anti-legacy/requirements/requirements_graph.json exists. If it does not, stop and run anti-legacy:graph-translator first — the deliverables render FROM the graph.

Then run the readiness gate — the authoritative precondition. Unlike a file-exists check it refuses on an incomplete or state-desynced pipeline (a confidence-less rule, coverage < 1.0, or an orphaned requirements-graph whose legacy-graph evidence is gone — ROOT A/B):

python3 .anti-legacy/run.py precheck deliverables

If it exits non-zero, STOP and resolve the named blockers. The Tier-A renders below call this same gate internally and will refuse on their own; --force is the explicit, loud override (only for a deliberate partial/preview run). The living logs (risk_log, decisions_log, evidence_log) are intentionally not gated — they must run on an incomplete pipeline to SURFACE its gaps.

Step 2: Render the deliverables

Run each render through the dispatcher, in this order. A render that must degrade (e.g. test-scripts on a stack it cannot script yet, or any Tier-A render with no blueprint) prints a note and still exits 0 — do not abort the suite on a soft note; collect it for the report.

python3 .anti-legacy/run.py prd
python3 .anti-legacy/run.py diagrams
python3 .anti-legacy/run.py test_plan
python3 .anti-legacy/run.py test_scripts
python3 .anti-legacy/run.py migration_plan
python3 .anti-legacy/run.py risk_log
python3 .anti-legacy/run.py decisions_log
python3 .anti-legacy/run.py evidence_log

A hard non-zero exit means a true precondition failure (no graph, or no manifest for the evidence log) — surface it and stop; do not paper over it.

Optional — richer output via parallel subagents. The renders above are deterministic and fast. Two deliverables benefit from agent reasoning on top of the deterministic skeleton: the PRD (narrative framing of each requirement) and the functional test scripts (concrete assertions fleshed out from each contract scenario). When you want that, dispatch anti-legacy:prd and anti-legacy:test-scripts as subagents — they have no ordering dependency on each other, so run them in parallel.

Step 3: Compile the index

python3 .anti-legacy/run.py deliverables_index

This writes + registers .anti-legacy/deliverables/README.md — the package table of contents: each deliverable's path, status, produced-at, and a present/absent receipt, plus the canonical expected set so anything not yet produced is named, not silently absent.

Step 4: Adversarial review (advisory — not a gate)

The renders are deterministic and trusting — they lay out what the data says. Before sharing the package, run the adversarial review that distrusts them: it dispatches one read-only critic subagent per deliverable, in parallel, to challenge each rendered file against the source data it was rendered from — hunting unsupported/optimistic claims, dropped §2 traceability, empty/weak sections, missing parity rules on numeric outputs, "done" that hides a gap, and divergence from the requirements graph.

Run anti-legacy:adversarial-review (see that skill for the critic micro-context and the parallel-dispatch protocol). It assembles the worklist via:

python3 .anti-legacy/run.py deliverable_review_worklist --json

then dispatches the critics and returns a structured per-deliverable verdict (findings[] with severity → PASS / REVISE / BLOCK) plus an aggregated package verdict.

This is advisory adversarial review, NOT a gate. It clears no gate, advances no phase, and registers no artifact. On REVISE/BLOCK, re-run the named producing deliverable to fix the render at its source and re-review, or proceed with an explicit, stated --force reason (mirroring the precheck override — never a silent skip). Collect the verdicts for the report in Step 5.

Step 5: Done-gate, then report

Done-gate (BLOCKING). Assert the index exists and is non-empty and at least one deliverable was produced:

python3 -c "import os,sys; \
idx='.anti-legacy/deliverables/README.md'; \
ok=os.path.isfile(idx) and os.path.getsize(idx)>0; \
sys.stderr.write('' if ok else 'deliverables done-gate FAILED: index missing/empty\n'); \
sys.exit(0 if ok else 1)"

If it fails, surface which renders errored and stop. This skill never calls manifest advance — it registers artifacts only.

Report to the user:

Deliverables package: .anti-legacy/deliverables/ (open README.md for the index)
Produced N / 9 and any deliverable that degraded (which input it was missing)
Adversarial review verdict (Step 4): the package verdict (PASS / REVISE / BLOCK) and any CRITICAL/MAJOR findings per deliverable — advisory, cleared no gate
Reminder: re-run risk-log, decisions-log, evidence-log at each gate to refresh the living deliverables

Output

.anti-legacy/deliverables/ — the full package (PRD, diagrams/, test-strategy, tests/, migration-plan + Jira CSV, risk/decisions/evidence logs)
.anti-legacy/deliverables/README.md — the registered index (deliverables-index)
Manifest: deliverable-* artifacts registered; phase unchanged
Adversarial-review verdicts (anti-legacy:adversarial-review, Step 4) — advisory, no artifact

Next step: address any REVISE/BLOCK from the adversarial review (re-run the named producing deliverable, then re-review), then share the package via git/fileshare and feed it into the review-packet / GATE_1 design review. Re-run the living logs at later gates.