• v1.8.0 137ae76c94

    v1.8.0
    All checks were successful
    CI / json-check (push) Successful in 6s
    CI / security (push) Successful in 8s
    Publish / detect-release (push) Successful in 5s
    CI / manifest (push) Successful in 6s
    Publish / tag (push) Has been skipped
    CI / lua-check (push) Successful in 11s
    Publish / promote-release (push) Has been skipped
    CI / sast (push) Successful in 39s
    Stable

    Gravity Bot released this 2026-02-17 00:08:22 +00:00 | 412 commits to master since this release

    v1.8.0 — Pre-Distribution Research

    Research-only release. All deliverables are OpenProject wiki artifacts — no code changes beyond the version bump.

    Research Completed

    • OP#648: AGENTS.md Standard — Evaluated the AAIF/Linux Foundation AGENTS.md standard. Gravity's skill system is a strict superset; recommendation is to generate AGENTS.md as a distribution artifact. Wiki: agents-dot-md-evaluation

    • OP#651: Agent-Agnostic Distribution — Surveyed 10 AI coding platforms' config locations. Recommendation: adapter via manifest targets array extension, not separate per-tool directories. Wiki: agent-agnostic-distribution-strategy

    • OP#655: Standalone CLI Extraction — Analyzed sync engine portability (70% portable). Recommendation: Go CLI with Bubble Tea TUI, CLI-first architecture with Neovim as thin wrapper. Wiki: standalone-cli-architecture

    • OP#656: Cross-Platform Support — Identified 5 critical breaks on macOS, 0 on WSL, 2-3 on containers. Priority: WSL (near-zero effort) > macOS > containers. Wiki: cross-platform-compatibility

    • OP#657: Editor Formatting Parity — All formatters are standalone CLIs. Parity via .editorconfig + tool configs + .vscode/settings.json + CI enforcement. Biome over Prettier for Cursor. Wiki: editor-formatting-parity

    • OP#652: Stakeholder Interface Options — Evaluated 6 interface options. Recommendation: WP comment-based approval with work_package:commented webhook (OP v16.2.0). Wiki: stakeholder-interface-options

    • OP#661: Sidecar RALPH Loops — Designed hybrid dispatcher (long-running Go) + ephemeral claude -p per event. Five-gate termination model, four-level budget hierarchy. Wiki: sidecar-automation-patterns

    • OP#743: Formal Skill Verification — Designed STD (Skill Test Document) format with worked example. 7 verification patterns evaluated (MBT, PBT, AI Red Teaming, BDD, DbC, fuzzing, metamorphic). Wiki: skill-verification-architecture

    What's Next

    These research artifacts feed directly into v2.0 — Distribution, which will implement the standalone CLI, agent-agnostic distribution, and cross-platform support.

    References

    Downloads