Skip to content

v1.1.0 - align with spec-kit conventions, cover the whole spec, safe re-runs

Latest
Latest

Choose a tag to compare

View all tags
@stn1slv stn1slv released this 09 Aug 09:43
v1.1.0
57908a3
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Verified
Learn about vigilant mode.

Breaking: the feature directory is now a required first argument.

/speckit.reconcile.run specs/007-invoice-settings "The /api/v1/settings endpoint now requires an org_id header"

It previously came from check-prerequisites.sh, which resolves whichever feature was worked on last. For a command that rewrites three files in place, that default was not worth its ambiguity.

Fixed against spec-kit conventions

Verified against the v0.16.1 templates, not inferred:

  • [P] is no longer emitted. It means "can run in parallel (different files, no dependencies)", not priority, and /speckit.implement reads it that way, so blocking remediation tasks were being written into tasks.md as safe to parallelize. It is also never stripped from tasks /speckit.tasks wrote.
  • Tasks are placed under the phase heading that already exists, rather than a new one built from the inline [USn] task tag, which is not a heading.
  • Spec edits target **Acceptance Scenarios** under each user story, not an "Acceptance Criteria" section with AC-XX IDs. Neither exists in spec-template.md.

Added

  • Every spec section is reachable. Functional Requirements, Edge Cases, Key Entities, Success Criteria and Assumptions, alongside Acceptance Scenarios. A gap report describing a changed capability previously had nowhere to land.
  • Scope modifiers are enforced. --spec-only, --plan-only and --tasks-only were declared and then never referenced again, so they had no effect on any edit.
  • Re-runs are safe. A [Sync: slug] key means a refined gap report updates what it already wrote instead of duplicating it, and a task /speckit.implement has marked [X] is never edited.
  • tasks.md is no longer created during the setup gate, before the clarification pause and the impact map that previews every edit.

Verification status

Reviewed across four rounds by two independent reviewers, but not yet exercised against a real project. This repository ships Markdown prompts and has no test suite. Feedback from a real run is welcome, particularly on the first archival of a fresh feature and on re-running the same gap report twice.

Assets 2
Loading