openPrompting v1.1.0

openPrompting 1.1.0

openPrompting 1.1.0 expands the source-backed catalog while keeping the V1 local CLI and configuration contracts stable.

Highlights

• Added sourced model guidance for OpenAI GPT-5.3-Codex and Anthropic Claude Opus 5.
• Added harness guidance for Cursor, GitHub Copilot CLI, OpenCode, and Aider, with model behavior kept separate from harness behavior.
• Added planning, research, testing, and migration task recipes with matching deterministic templates.
• Added task/template integrity validation, dynamic catalog checks, and packed-package journeys for every active task.
• Added optional display_name metadata so setup, guide, and compare output can be easier to scan without changing stable IDs.
• Deepened current GPT-6 Astra, Claude Sonnet 5, and Gemini 3.8 Flash guidance from first-party documentation.

Compatibility

The executable remains openprompting, and the command surface remains help, setup, guide, new, doctor, and compare. Configuration continues to use version: 1; existing V1 configuration values and IDs resolve unchanged. The evidence vocabulary remains official, tested, community, and legacy.

Validation

The release gate covers lint, type checking, the full test suite, build, knowledge validation, README verification, package dry-run, and installation of the packed artifact into an isolated consumer. The packed consumer exercises the original V1 journey, new model and harness selectors, and both default and explicit-profile rendering for every active task.

Operation and privacy

Normal CLI operation remains local and deterministic. This release adds no runtime model calls, telemetry, hosted backend, workflow execution, or automatic network access.

Deferred

Remote knowledge updates, native harness adapters, handoffs and workflows, prompt optimization, hosted documentation, benchmarking, and additional harnesses remain outside this release.