name: control-center description: Centralized command and control for AI skills and context management. Use this skill to organize, index, and orchestrate multiple skills across different projects. It acts as a "Mission Control" to identify the best tool for a task, manage shared context between skills, and reduce "context pollution" by selectively activating only necessary resources.
Control Center (Mission Control for AI Agents)
This skill provides a centralized logic layer to manage multiple tools, skills, and project contexts efficiently.
Core Capabilities
- Skill Discovery: Indexes all available skills in
.gemini/skills/and.agents/skills/. - Context Orchestration: Decides which skill to activate based on the complexity and scope of a task.
- Efficiency Optimization: Prevents "context flooding" by loading only relevant
SKILL.mdcontent and documentation. - Tool Chaining: Designs workflows that involve multiple skills (e.g.,
agent-browser+pdf+xlsx).
Standard Operating Procedures
Step 1: Capability Mapping
When a complex task is received, identify all required capabilities:
- Search:
search-skill,google_web_search - Automation:
agent-browser,control-center - Analysis:
codebase_investigator,skill-vetter - Production:
pdf,xlsx,docx
Step 2: Workflow Design
Instead of running tools one-by-one, design a sequence:
- Phase A (Research): Use
agent-browserorsearch-skill. - Phase B (Verification): Use
skill-vetter(if installing new tools). - Phase C (Execution): Use project-specific tools (e.g.,
xlsx). - Phase D (Documentation): Record results to KM.
Step 3: Context Governance
- Is the context too large? If so, use
control-centerto summarize previous steps and clear irrelevant history. - Is the task multi-turn? Maintain a "Session State" to track progress across multiple skill calls.
Reporting Format
When managing a complex workflow, provide a brief "Mission Briefing":
- Mission Name: [Clear goal]
- Activated Tools: [Skill A, Skill B]
- Current Status: [Researching/Executing/Validating]
- Next Milestone: [Clear next step]
Best Practices
- Prefer Specialization: If a specialized skill exists (e.g.,
pdf), always use it instead of a generalist tool. - Enforce Security: Always run
skill-vetterthroughcontrol-centerwhenever a new source is involved. - Document Everything: Use the KM workflow to ensure every "Mission" leaves a permanent record.