oasis7

Use when any oasis7 user request starts and must bind the canonical worktree, GitHub-backed task truth, and owner before routing.

Use when a branch is about to create a PR and needs fresh involved-role review.

Use when implementation is verified and the branch must proceed through review, PR, merge, and cleanup.

Use when a task has written repo truth and implementation should proceed step by step with evidence.

Use when about to claim a task is complete, tests passed, a branch is ready for PR, or a PR is ready to merge. Requires fresh verification, direct output inspection, and use of `./scripts/pm/claim-ready.sh` when the claim can be mapped to one verification command.

Use when a bound oasis7 task needs the next repo-owned workflow phase selected.

Use when assessing or developing the blocked production-supervisor target for a bound oasis7 task spanning multiple lifecycle phases.

Use when creating or moving a repo-owned skill surface under .agents/skills or skills, replacing upstream guidance, or editing trigger wording, helper references, or governance links.

Use when automating browser interaction with the local agent-browser CLI, including navigation, forms, screenshots, extraction, authentication, responsive checks, or visual diff workflows.

Use when drafting or adapting oasis7 liveops, community, channel, release, or marketing copy that needs format, headline, SEO, CTA, or campaign guidance.

Use when planning, extending, or drafting a Chinese long-form, world-heavy story or lore bible that needs reusable world rules, character registries, timeline control, chapter cards, consistency checks, and repo-tracked writeback.

Use when planning a new game architecture, major game feature, refactor, or technical design document that needs structured requirement, technical design, and implementation-plan outputs.

Use when an oasis7 gameplay, product, agent, visual, or liveops slice needs theory-backed framing for player motivation, core loops, progression, balance, or experience design.

Use when an oasis7 change creates or modifies player-visible interaction behavior, including player-facing flows, input/control behavior, UI state feedback, error recovery, accessibility implications, or game UX heuristics for viewer, web, launcher, or gameplay screens.

Use when an oasis7 change creates or modifies visual/player-visible game UI content, including screen hierarchy, readability, contrast, composition, density, motion emphasis, screenshot review criteria, or visual reference workflows for viewer, web, launcher, or gameplay screens.

Use when implementing or refining concrete gameplay mechanics, feedback loops, combat, economy, progression, movement, event interactions, or balance iteration.

Use when generating or editing raster images through GPT Image 2 / ChatGPT Images 2.0 using the user's ChatGPT subscription.

Use when editing Chinese text to reduce AI-written patterns, improve natural voice, remove promotional phrasing, and preserve the author intent without adding unsupported meaning.

Use when designing or evaluating game levels, pacing, difficulty progression, spatial layout, environmental storytelling, encounters, or traversal flow.

Use when optimizing game memory usage, object pooling, allocation behavior, garbage collection, memory profiling, asset streaming, or platform memory budgets.

Use when optimizing game performance, profiling frame rate, CPU/GPU bottlenecks, loading time, scalability, or multi-platform runtime behavior.

Use when creating, tuning, optimizing, pooling, or troubleshooting particle effects, visual effects recipes, GPU particles, or engine-specific VFX tooling.

Use when generating or improving Product Requirements Documents with executive summaries, user stories, technical specifications, acceptance criteria, and risk analysis.

Use when an oasis7 runtime slice needs to analyze or design authority, state consistency, reconciliation, prediction, interpolation, lag handling, or multiplayer synchronization behavior.

Use when reviewing or optimizing XiaoHongShu / RedNote note content for title quality, keywords, sensitivity risk, commercialization level, and engagement potential.

Use when the task is still ambiguous, needs scope decomposition or option comparison, or is inherently visual enough to justify a visual companion before implementation. Converts upstream brainstorming into a repo-owned, non-mandatory ideation step.

Use when a behavior-changing task needs failing tests first, RED/GREEN handoff, or automated behavior gates before implementation.

Use when a PR receives review comments or when a user asks to handle review feedback. Verifies each comment against code and repo truth, applies minimal valid fixes, and keeps thread resolution separate from merge readiness.

Use when you hit a bug, failing test, broken script, unexpected diff, or behavior regression. Focuses on reproduction, narrowing the failure surface, validating hypotheses, and only then patching the root cause.