メニュー
Unavailable archived skill. Do not use to initialize, continue, repair, or manage long-running projects across sessions; retained only for historical reference and postmortem analysis.
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
SOC 職業分類に基づく
Use when working on Gentoo ebuilds, overlays, eclasses, Manifest files, metadata.xml, pkgdev, pkgcheck, ebuild QA, version bumps, new packages, package reviews, or Gentoo package pull requests.
Evidence-backed visual repository analysis for understanding codebases, creating concise Mermaid architecture/flow diagrams, mapping components and runtime/data paths, identifying hotspots, and producing repository-informed improvement opportunities. Use when the user asks to analyze a repository, understand a codebase, generate architecture diagrams, create a visual repo report, find high-value improvement directions, or prepare context for later design/planning work.
Use across bugs, code changes, reviews, architecture, product decisions, writing, research, and general problem solving when the user presents a visible issue, example, proposal, or decision. MUST apply Locez Lens as a lightweight scope and judgment calibration before answering or changing anything.
Use when creating or refactoring a skill for a complex multi-step task, especially when an existing skill has an overloaded main prompt, unclear sub-agent responsibilities, repeated human confirmations, missing intermediate artifacts, weak validation, or a need to turn domain-specific work such as data cleaning into a structured workflow skill.
Use when bumping, removing, or reconciling nvchecker-tracked packages in locez-overlay, especially when deciding whether a package is release-managed, live-only, or no longer maintained
Use when handling Portage conflicts, @world upgrades, /etc/portage cleanup, binpkg tradeoffs, or deciding where Gentoo maintenance rules and local host preferences should live
| name | project-continuity |
| description | Unavailable archived skill. Do not use to initialize, continue, repair, or manage long-running projects across sessions; retained only for historical reference and postmortem analysis. |
This skill is unavailable and must not be used for active work.
Do not use it to initialize, continue, repair, or manage project state across sessions. Do not emit /goal $project-continuity as a next command. Do not create or update continuity artifacts from this skill.
The skill directory is retained only as an archive for historical review, migration analysis, or postmortem work. Its former references, templates, and workflow assets remain on disk, but they are inactive.
Experiments showed that this workflow depends too heavily on durable state and strict state updates. User instructions naturally interrupt and redirect sessions, which can prevent the state files from staying coherent. Once continuity breaks, the stored project picture can become seriously misleading for a fresh session.
The workflow also depends too much on model capability. If the model's reasoning quality drops, the lease boundaries, state repair, and handoff discipline degrade quickly, producing poor delivery quality and compounding bad state.
See the repository decision records:
docs/specs/2026-05-29-project-continuity-unavailable.mddocs/specs/2026-05-29-project-continuity-unavailable.zh-CN.md