用 Codex 或 Claude 帮你安装 复制这段 Prompt,粘贴到 Codex、Claude 或其他助手里,让它检查 Skill 页面并帮你完成安装。
直接命令不会经过审查 Prompt;运行前请先检查来源。
npx skills add https://github.com/applicate2628/Orchestrarium --skill external-brigade命令会保持在同一行。复制前请横向滚动并检查完整内容。
想先保存到本地?可下载 SkillsMP 当前能够提供的文件。
基于 SOC 职业分类
正在显示 SKILL.md
| name | external-brigade |
| description | External brigade: launch bounded parallel helper lanes. |
Use this skill when the main Gemini session needs one bounded plan for multiple external helper launches instead of scattering the orchestration across several separate notes.
$lead or $product-manager.Each brigade item must name:
consultant, external-worker, or external-reviewernone for pure advisoryautoOne brigade item equals one helper instance, one admitted artifact, and one gate.
.gemini/.agents-mode.yaml before trusting any flags..gemini/.agents-mode.yaml is missing, read local legacy .gemini/.agents-mode as compatibility input only; if both local files are missing, fall back through pack-local global ~/.gemini/.agents-mode.yaml, pack-local global legacy ~/.gemini/.agents-mode, then the shared cross-pack global ~/.agents-mode.yaml (alongside ~/.claude.json), before applying built-in defaults. Each key resolves to the highest layer that defines it; layers compose, they do not replace each other wholesale. Normalize whichever file supplied the effective config into the canonical .yaml path in the same scope and do not recreate any legacy file.consultantModeparallelModeexternalProviderexternalPriorityProfilereserveResolverexternalPriorityProfilesexternalOpinionCountsexternalCodexWorkdirModeexternalClaudeWorkdirModeexternalModelModeexternalCodexProfileparallelMode as the general helper fan-out rule. Brigade launch is an explicit bounded overlay on top of that rule, not a second general concurrency model.externalOpinionCounts scoped to same-lane distinct-opinion requirements. It does not cap how many same-provider brigade items may run in parallel across different disjoint lanes or slices.2+ same-lane opinions, satisfy that distinct-provider requirement first or fail that item closed.externalCodexProfile when Codex is resolved; default inherits externalModelMode; gpt-5.6-sol-max requests model gpt-5.6-sol with model_reasoning_effort = "max" for higher-complexity/hard lanes; gpt-5.6-terra selects the balanced Codex model tier (a distinct model, model_reasoning_effort = "high", not an effort downgrade) and requires installed-runtime verification; gpt-5.6-sol-xhigh (shipped as default in the Codex/Claude packs) pins model gpt-5.6-sol with model_reasoning_effort = "xhigh" regardless of externalModelMode. Honor externalModelMode before other provider-specific transport knobs. On shipped production paths that means Codex/Claude behavior only; explicit Gemini or Qwen runs remain manual example or compatibility paths rather than a separate production fallback policy.When a brigade review/advisory item resolves to reserve, bind that symbolic supplemental candidate through reserveResolver. Worker-side brigade items must not use reserve.
Example-only provider rule:
WEAK MODEL / NOT RECOMMENDED.auto profiles.gemini or qwen explicitly for a manual example or compatibility run.Return one brigade report with:
PASS | REVISE | BLOCKED:dependencyThe launch table must keep these columns explicit:
| Item | Execution role | Assigned / replaced internal role | Requested provider | Resolved provider | Scope | Result |
|---|---|---|---|---|---|---|
<item> | <role> | <role or none> | `<internal | codex | claude | gemini |
PASS only when every required brigade item produced its required artifact.REVISE when at least one required item returned REVISE and no required item is blocked.BLOCKED:dependency when a required item cannot run, when a same-lane distinct-opinion count cannot be satisfied, or when the requested brigade is structurally invalid.