Prove PowerShell script block data safely

[Aaronontheweb]

Summary

• keep script blocks opaque only when a constrained PowerShell baseline and bounded command-resolution state prove the receiver consumes data
• retain unknown execution regions for default state, exact module-qualified-looking aliases, and canonical alias-target mutation
• preserve unrelated-name precision and fresh child-runspace reset semantics
• promote proved data, unknown receiver, local Invoke-Command, and both alias-shadowing attacks into the generated live-pwsh corpus
• correct the OpenSpec, PowerShell spec, design corpus, task ledger, and implementation plan

Security boundary

PowerShell permits aliases named Microsoft.PowerShell.Utility\Write-Output, and echo resolves through canonical Write-Output. Both can execute a script block after mutation. The analyzer now matches exact mutations against authored and canonical identities before removing speculative execution regions. Unknown or unbounded state remains visible and incomplete.

Verification

• adversarial review: PASS after two reproduced shadowing blockers were fixed
• dotnet build -c Release --no-restore
• dotnet test -c Release --no-restore (2575 passed)
• focused parser/catalog/corpus/live-pwsh/PII suite (985 passed; reviewer independently ran 1069)
• pwsh ./scripts/Add-FileHeaders.ps1 -Verify
• openspec validate v0-3-structured-shell-analysis --strict
• slopwatch analyze -d . --no-baseline --fail-on warning (0 findings)
• generated PowerShell corpus round-trip (422 entries, exact directory diff)