Plan review

Self-building AI code generation engine that generates application flows instead of implementing them. AGPL-3.0.From the repository description

Install

npx -y skills add lubochka/xiigen-mvp-engine --skill plan-review

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

2 things to look at

19 days oldThe repository was created 19 days ago. New is not bad, but a brand new repository carrying a familiar-sounding name is the shape a typosquat arrives in, and there has been no time for anyone else to find a problem with it.
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.

1.9 KB, 325 tokens by cl100k_base, as published. Nobody here has run it

Use this before executing a plan or handing it to another agent so weak assumptions are caught while they are cheap to fix.

Invoke after drafting a plan, before implementation, before multi-agent handoff, or when a plan changes due to new evidence.

Check that every user requirement appears in the plan or is explicitly ruled out with a reason.
Confirm the plan uses only authorized files and avoids prohibited areas.
Verify ordering: reconnaissance before edits, contract changes before dependents, tests/docs after behavior changes.
Look for missing gates, rollback options, and ambiguous completion language.
Record review findings as required fixes, optional improvements, or accepted risks.

For XIIGen plans, ensure service/API work includes DataProcessResult<T> outcomes, Promise/AbortSignal behavior where locally present, NestJS provider or injection-token wiring, existing config helpers/defaults, structured logging, and test impact where relevant.
For search or persistence plans, confirm buildSearchFilter/buildSearchFilterFlat behavior and Record<string, unknown>/unknown payload assumptions are not replaced by fixed schemas.

Do not approve a plan because it sounds plausible; bind it to files, commands, and evidence.
Do not reopen settled user constraints unless the user asks.
Do not import source architecture, algorithms, design patterns, training, DPO, source domains, source classes, or source-specific paths.

The plan is executable, scoped, evidence-backed, and any remaining risk is named before work proceeds.

Read from the repository

Just SKILL.md. No reference files, no scripts.