mss-module-review

mss-module-review

Use this skill as an MSS/module review checklist and delegation layer.

Deterministic module diagnostics and flow evidence are owned by plaited-context:

• skills/plaited-context/scripts/context.ts
• skills/plaited-context/scripts/module-patterns.ts
• skills/plaited-context/scripts/module-flow.ts

This skill no longer owns executable module-analysis scripts.

Skill Coordination

Use the typescript-lsp skill when symbol identity, references, definitions, or alias-heavy helper flow is ambiguous.

Notes:

• scanFile / workspace_scan are import/export indexing only.
• symbols, references, and definition are the relevant TypeScript LSP probes for call-edge confidence.
• Type-aware probes are review evidence, not a replacement for deterministic rules.
• Do not claim there is no useful LSP support.

Required Review Commands

Use plaited-context for context assembly:

bun skills/plaited-context/scripts/context.ts '{"task":"review module actor diagnostics","mode":"review","paths":["<module-files>"]}'

Run hard gate:

bun skills/plaited-context/scripts/module-patterns.ts '{"files":["<module-files>"]}'

Run required review evidence:

bun skills/plaited-context/scripts/module-flow.ts '{"files":["<module-files>"],"format":"json"}'
bun skills/plaited-context/scripts/module-flow.ts '{"files":["<module-files>"],"format":"mermaid"}'

When ambiguity exists, run typescript-lsp probes:

bun skills/typescript-lsp/scripts/run.ts '{"file":"<module-file>","operations":[{"type":"symbols"}]}'
bun skills/typescript-lsp/scripts/run.ts '{"file":"<module-file>","operations":[{"type":"references","line":<line>,"character":<character>}]}'
bun skills/typescript-lsp/scripts/run.ts '{"file":"<module-file>","operations":[{"type":"definition","line":<line>,"character":<character>}]}'

Optional: include DB recording in plaited-context commands with "record": true and a configured "dbPath" when findings/evidence should be persisted.

Review Checklist

• flat src/modules/*.ts file boundary
• src/modules/*.ts means core runtime module actors, not user-facing module programs
• user/developer-authored proposals remain module programs until admitted
• src/modules.ts remains the public modules boundary
• forbidden nested module implementation folders
• projection views are governed audience views of one source module program unless separate state/provenance is explicit
• declared access requests are asks, not grants, and remain metadata unless a later grant path exists
• external boundary vs internal useExtension(...) feedback/control parsing
• no internal handler safeParse(...) silent drops
• no internal handler ZodError recovery
• actor/runtime diagnostics use reportSnapshot(...)
• transport diagnostics stay at external transport boundaries
• no synthetic diagnostic events unless there is a real protocol consumer
• no hidden Bun/HTTP/WebSocket/MCP/A2A servers in semantic boundary actors
• source/provenance is not rewritten to fake approval
• supervisor/projection/transport ownership is respected

Final Handoff Format

Report:

• module files reviewed
• hard gate output
• graph JSON summary
• Mermaid snippet
• TypeScript LSP probes run, if any
• architectural findings
• known limitations

Known Limitations

The flow renderer is AST-first and local to each useExtension(...) callback. It handles direct local helper chains and expression-bodied helpers, but may still miss alias-heavy, destructured, reassigned, imported, or cross-file helper indirection. Use the typescript-lsp skill to verify symbol identity when the graph is sparse or ambiguous.