islee23520

Lazy-load for LUX Unity bridge and MCP game-development operations: status, compile, tests, logs, launch, screenshots, hierarchy, dynamic code, record/replay, and one-loop MCP workflows.

Lazy-load only when this workflow is explicitly needed. Use for durable technical choices. Output a short ADR: Context, Decision, Alternatives, Consequences, Verification, Owner. Prefer existing LUX boundaries: gateway owns CLI/server, bridge owns Unity protocol, Skills owns passive workflows. Reject duplicate state, hidden fallback, and broad rewrites without evidence.

Lazy-load only when this workflow is explicitly needed. Use to review an existing design or implementation for boundary drift. Check SSoT, responsibility split, protocol consistency, atomicity, idempotency, and observable failures. Return ranked findings with file references, risk level, and the smallest corrective action.

Lazy-load only when this workflow is explicitly needed. Use when a change risks architectural drift. Enforce: .lux is SSoT; clear subsystem ownership; consistent schemas; atomic operations; idempotent retries; no silent fallback. Report violations and concrete repair steps.

Lazy-load only when this workflow is explicitly needed. Use for high-risk product, automation, or ethical decisions. Run: clarify intent; quantify evidence; check user/data/IP/fairness risks; define stop and rollback conditions; record approval. Verdicts: PASS, REVIEW, or REJECT.

Lazy-load only when this workflow is explicitly needed. Use to turn observed failures into actionable reports. Capture summary, environment, exact steps, expected result, actual result, logs, suspected subsystem, severity, reproducibility, and minimal next diagnostic. Do not invent fallback data.

Lazy-load only when this workflow is explicitly needed. Use to classify and route defects. Determine impact, scope, likely owner, regression window, blocking status, and verification path. Prefer quick reproduction and narrow hypotheses before implementation.

Lazy-load only when this workflow is explicitly needed. Use to write release notes from verified changes. Group by Added, Changed, Fixed, Removed, Security, and Internal. Mention user-visible impact first. Exclude unverified claims and noisy implementation detail.

Execute a Prometheus work plan in GrokBuild with `/goal` state, Boulder state, evidence ledger updates, worktree discipline, parallel subagents, and explicit continuation. Use after planning when the user says start work, execute plan, continue plan, resume plan, or asks to run a .omo/plans plan.

Goal-like loop that uses ultrawork mode to decompose work into systematic, evidence-bound steps.

MUST USE for planning before coding: 5+ steps, ambiguous scope, multiple modules, architecture decisions, a vague 'just make it good / figure out what to build' brief, or any request to plan, interview, or break work down. Explore-first planning consultant (Prometheus) that grounds in the codebase, asks only the forks exploration cannot resolve - or researches them to best practice when the intent is fuzzy - waits for explicit approval, then writes ONE decision-complete work plan a worker executes with zero further interview. Triggers: ulw-plan, plan this, make a plan, plan before coding, interview me, break this down, start planning, plan mode, just make it good, figure out what to build.

Execute a Prometheus work plan in GrokBuild with `/goal` state, Boulder state, evidence ledger updates, worktree discipline, parallel subagents, and explicit continuation. Use after planning when the user says start work, execute plan, continue plan, resume plan, or asks to run a .omo/plans plan.

Goal-like loop that uses ultrawork mode to decompose work into systematic, evidence-bound steps.

Persistent code-mode kernels (JS/Python) for incremental computation — omp/senpi eval analog on GrokBuild via lfg MCP. Use when multi-step code execution, data exploration, or stateful REPL beats one-shot bash.

Post-implementation review orchestrator. Launches 5 parallel background sub-agents: Oracle (goal/constraint verification), Oracle (code quality), Oracle (security), unspecified-high (hands-on QA execution), unspecified-high (context mining from GitHub/git/Slack/Notion). All must pass for review to pass. MUST USE after completing any significant implementation work. Triggers: 'review work', 'review my work', 'review changes', 'QA my work', 'verify implementation', 'check my work', 'validate changes', 'post-implementation review'.

Read Claude Code plugins, skills, project auto-memory, and the durable Grok↔Claude message bridge. Use when the user asks what Claude plugins/skills are installed, to inspect Claude MEMORY.md / project memory, or to send/read messages between Grok/lfg and Claude Code (mailbox or live `claude -p` ask). Triggers: claude plugins, claude skills, claude memory, Claude Code inventory, read claude skill, marketplace plugins, message claude, talk to claude, claude bridge, lfg claude ask.

Use only for general codex-xai-oauth/xAI Grok plugin overview, setup orientation, or multi-tool requests when no dedicated xai_* skill is a better fit.

Use when the user asks how to log in, authenticate, configure OAuth, set up XAI_API_KEY, or fix missing xAI/Grok credentials for codex-xai-oauth.

Use when the user asks to generate, draft, answer, rewrite, summarize, or reason with xAI/Grok text generation through xai_generate_text.

Use when the user asks to generate, create, make, or save images with xAI, Grok Imagine, xai_image_generate, or codex-xai-oauth image artifacts.

Use when the user asks to check xAI, Grok, codex-xai-oauth, OAuth, API-key, or credential status without exposing secrets.

Use when the user asks to generate speech, text-to-speech, audio narration, voice output, or TTS with xAI/Grok through xai_tts.

Use when the user asks to generate video, Grok Imagine video, text-to-video, image-to-video, or xAI video through xai_video_generate.

Use when the user asks xAI/Grok to search the web, find current information, answer with citations, or use server-side web_search.

Lazy-load for LUX Unity bridge and MCP game-development operations: status, compile, tests, logs, launch, screenshots, hierarchy, dynamic code, record/replay, and one-loop MCP workflows.