| name | anti-legacy:expert |
| description | 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.
|
anti-legacy-expert — pipeline internals & shared core
This skill is the subject-matter reference for the anti-legacy pipeline's shared
Python core, and it physically hosts that core at scripts/antilegacy_core/. The
library ships with the portable bundle because the skills standard delivers a
skill's entire directory subtree (symlink or --copy).
The shared core (antilegacy_core)
All cross-cutting pipeline logic lives in one namespaced, collision-proof package
(no more bare import coverage shadowing stdlib / pytest-cov):
| Module | Responsibility |
|---|
estate | The wicked-estate engine seam — resolve binary, index/query/source/annotate the code graph. |
coverage | Resolved-or-flagged coverage over the graph + annotations.jsonl. |
extract | Business-rule extraction from source slices. |
vocabulary | Glossary / domain-term mining + curation. |
domain_graph | The requirements-graph builder (§I5). |
normalizer | Code-graph → draft requirements scaffold. |
comparator | Graph diff / round-trip comparison. |
manifest | The pipeline state machine — phases, gates, audit. |
validator | The build / semantic / UAT evidence verifier. |
planner | Task-list planning helpers. |
preflight() (in __init__) | Host-agnostic readiness check (engine, deps, workspace). |
The pipeline JSON schemas ship as package data under antilegacy_core/schemas/
and are read via importlib.resources — one source of truth, versioned with the
code.
How it is invoked
Skills never import the core directly. They dispatch through the workspace seam:
python3 .anti-legacy/run.py <stem> <args...>
run.py resolves <stem> to a library module and runs
python -m antilegacy_core.<stem>, with the package on PYTHONPATH. Workspace
state (config, annotations, graphs) is anchored on the current working directory —
never on the package's __file__ — so the core behaves identically whether the
bundle was installed as a plugin (whole repo) or via the skills standard (flat
per-skill dirs).
Readiness
python3 .anti-legacy/run.py preflight
Reports the engine (wicked-estate ≥ 0.5.1), Python deps, and workspace state, with
the exact remediation for anything missing. Fails fast — never silently degrades.
Extending the core
Add a module under scripts/antilegacy_core/. If it imports another core module,
it belongs here (the cross-importing core stays together). Single-consumer leaf
logic belongs in its owning phase skill's scripts/, not here.
