Skip to main content

dev-team

On-demand spawning of multiple engineering agents in parallel for large-context work. Use when a task is too big for a single agent and the work can be split into independent streams.

معلومات المصدر

المستودع
evolution-foundation/evo-nexus
آخر نشاط في المصدر
١٠ أبريل ٢٠٢٦ في ١٦:٢٤
لغة SKILL.md المكتشفة
الإنجليزية
النجوم
٥٤٧
التفرعات
١٨١

خيارات التثبيت

يُحدَّد Prompt الذي يراجع المصدر أولًا بشكل افتراضي. يمكنك التبديل إلى أمر مباشر أو تنزيل نسخة محلية.

مراجعة ملفات المصدر

اقرأ SKILL.md وأي ملفات مرافقة يعرضها SkillsMP قبل أن تقرر التثبيت.

عرض SKILL.md

SKILL.md
تعليمات المصدر · معاينة للقراءة فقط
name
dev-team
description
On-demand spawning of multiple engineering agents in parallel for large-context work. Use when a task is too big for a single agent and the work can be split into independent streams.
# Dev Team Derived from oh-my-claudecode (MIT, Yeachan Heo). Adapted for the EvoNexus Engineering Layer. On-demand parallel agent spawning. When a task is too big for one agent and the work can be split into independent streams, spin up multiple engineering agents working in parallel. ## Use When - Multi-file refactor across 10+ files in different modules - Codebase audit that benefits from parallel exploration - Performance investigation across multiple subsystems - Documentation pass across many files ## Do Not Use When - Sequential dependencies (B needs A's output) - Single-file change - Task fits within a single agent's context ## Workflow 1. **Decompose** the task into N independent streams 2. **Assign** each stream to the most appropriate agent (apex / bolt / lens / etc.) 3. **Spawn** agents in parallel via Task tool 4. **Collect** results as they return 5. **Synthesize** into a unified output 6. Save to `workspace/development/research/[C]team-{topic}-{date}.md` ## Streams Example For "audit the entire auth module": - **Stream 1:** `@scout-explorer` → map all auth files - **Stream 2:** `@vault-security` → security audit - **Stream 3:** `@lens-reviewer` → code quality review - **Stream 4:** `@grid-tester` → test coverage analysis - **Stream 5:** `@apex-architect` → architecture analysis All spawn in parallel, results combine into one report. ## Output ```markdown ## Team Investigation — {topic} ### Streams 1. {stream 1 result summary} 2. {stream 2 result summary} ... ### Cross-stream Findings [Insights that emerged from combining results] ### Recommendation [Unified next step] ``` ## Pairs With - Any of the 19 engineering agents (you pick which to spawn) - `dev-autopilot` (which can spawn its own team internally) - `@compass-planner` (often the consumer of team output)
عرض على GitHub