Skip to content

Harden structured Command operations across provider capabilities #44

Closed

Description

@triggeredcode

Goal

Preserve Lucid's checked Ops → BlockPatch safety model across providers with different structured-output behavior.

Depends on

Scope

  • Classify structured output as native, prompted-and-validated fallback, or unavailable.
  • Validate every response with the existing schema and compiler.
  • Preserve repair limits, stale-target checks, destructive confirmation, single-transaction undo, and unaffected blocks.
  • Prevent unsafe Command execution when a provider/model cannot meet the required contract.

Acceptance criteria

  • Adapter contract tests cover valid, malformed, partial, and schema-violating responses.
  • A capability gap produces an explicit UI state rather than a silent downgrade.
  • Existing Command evals run per supported provider configuration.
  • This issue extends the current operation system; it does not replace it.

Activity

  1. triggeredcode commented on Oct 19, 2026

    @triggeredcode
    OwnerAuthor

    Claimed. Implemented on feat/provider-context-roadmap in commit f2b571d and included in draft PR #46. Full CI passes (1,505 tests, 22 skipped), the Electron build passes, and Command/Settings E2E scenarios pass. Native, prompted-and-validated, and unavailable capability paths are explicitly tested.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

Labels

No labels
No labels

Projects

No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions