Lyra step 2.5 architecture deepening
Skill lyra-engine/lyra-step-2.5-architecture-deepening/skills/lyra-step-2.5-architecture-deepening
Thin lyra-step wrapper around `improve-codebase-architecture`. Surfaces deepening opportunities — turns shallow modules into deep, gray-box, AI-navigable modules — so the codebase is ready before contracts (11.5) and waves (15) are written. Phase B. Use after Step 2 (new projects) or right after /lyra-step-retrofit (existing projects), and as a recurring pass during Step 19 maintenance. Deutsch: "architektur vertiefen", "deep modules vor dem bauen", "codebase AI-ready machen".From its SKILL.md
npx -y skills add lyra-engine/lyra-step-2.5-architecture-deepening --skill lyra-step-2.5-architecture-deepeningAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
One thing to look at
- 0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
SKILL.md
3.6 KB, 716 tokens by cl100k_base, as published. Nobody here has run it
lyra-step-2.5-architecture-deepening
Thin lyra-step wrapper. Delegates the actual analysis to the improve-codebase-architecture skill. This wrapper only places that work in the Sigma/lyra-step sequence and records it in the workspace state.
Doctrine: Pocock "deep modules" (Matt Pocock — "Your codebase isn't ready for AI", YouTube 2026). A tangled shallow-module codebase makes interface contracts (11.5) impossible to write cleanly — deepen first.
Where it sits
Step 2 Architecture ─┬─> 2.5 deepening (new project: optional polish)
/lyra-step-retrofit ─┘ (existing project: REQUIRED before 11.5/15)
│
▼
11 PRD → 11.5 contracts → 15 build
Step 19 Maintenance ──> 2.5 deepening (recurring refactor pass)
When the user invokes this
- "/lyra-step-2.5-architecture-deepening [<workspace-path>]"
- "architektur vertiefen", "deep modules", "vor dem bauen aufräumen"
- automatically recommended by
/lyra-step-retrofit(part of the commercial lyra-step framework — not required) for existing projects with shallow-module sprawl
Workspace resolution
Arg 1 → workspace path, else pwd. Verify <workspace>/sigma/.state.json exists (warn if not — the underlying skill still runs against the code repo). Reads CONTEXT.md + docs/adr/ if present (the underlying skill uses domain language from there).
Process
- Run the
improve-codebase-architectureskill against the code repo. - Capture the report yourself (the wrapper's job, not the engine's): as soon as the engine has presented its numbered candidate list, write it to
<workspace>/sigma/step-2.5-architecture-deepening/output.md— even if the user explores no candidate further. If the user does enter the grilling loop, append the crystallized decisions to the same file. - Surface any proposal that changes a public interface as a
**Decision:**marker (it affects the contracts 11.5 will write). - Recommend: apply high-confidence deepenings BEFORE running 11.5, so contracts lock the new (deep) interfaces, not the old shallow ones.
Outputs
<workspace>/sigma/step-2.5-architecture-deepening/output.md— the deepening report (proposals, target interfaces, before/after module map).- Decisions appended to STATUS.md "Open Decisions" (if a Sigma workspace is present; standalone runs just keep the report).
STATUS.md / .state.json update
Sigma workspaces only — skip both files in standalone runs. Contract: in .state.json, set steps["2.5"].status to "done" (report written, no open decisions) or "partial" (report written, decisions open). In STATUS.md, append one line per open decision under the existing ## Open Decisions heading, formatted - [step-2.5] <decision>; never rewrite other sections. No emoji. Audit trail stays in the planning workspace; the refactors themselves land in the code repo.
What ships with it
Read from the repository
Just SKILL.md. No reference files, no scripts.