AgentSkillsCN

workspace-orchestrator-adapter

根据工作空间的上下文环境,结合用户长期使用过程中的提示行为,动态调整并优化技能编排策略。

SKILL.md
--- frontmatter
name: workspace-orchestrator-adapter
description: Adapts orchestration based on workspace context AND observed user prompt behavior over time.
triggers: [continuous, behavior-pattern-detected]
outputs: [orchestration-adjustments, preference-model]
depends_on: [user-intent-pattern-analyzer]

Workspace Orchestrator Adapter

Purpose

Dynamically adapts the orchestrator based on workspace context, project maturity, and observed user behavior patterns. The goal is to reduce friction, anticipate user intent, and optimize skill efficiency.


What This Skill Observes

Workspace Context

  • Project type (new, mature, legacy)
  • Technology stack
  • Team size and structure
  • Release frequency

User Behavior (passive observation)

  • Repeated prompt structures
  • Repeated clarification patterns
  • Repeated corrections by user
  • Frequency of user intervention
  • User tolerance to AI assumptions

User Preferences (inferred)

  • Speed vs safety preference
  • Detail vs summary preference
  • Automation vs control preference

Adaptation Actions

ObservationAdaptation
User always approves quicklyReduce confirmation prompts
User often corrects outputIncrease validation steps
User prefers detailed plansExpand planning output
User skips certain skillsDe-prioritize those skills
High-risk projectEnforce stricter gates
Rapid iteration phaseStreamline for speed

Preference Model

yaml
user_preferences:
  ai_autonomy: high  # low | medium | high
  assumption_tolerance: medium
  explanation_level: minimal  # minimal | normal | detailed
  speed_vs_safety: balanced  # speed | balanced | safety
  
workflow_adjustments:
  skip_low_risk_confirmations: true
  expand_planning_output: false
  require_explicit_approval: [security, compliance]
  auto_run_tests: true

Hard Constraints (Never Bypassed)

  • Spec governance requirements
  • Behavior contracts
  • Security skill checks
  • Compliance validations
  • Human approval for HIGH-risk changes

Integration

  • Receives from: user-intent-pattern-analyzer
  • Modifies: orchestrated-development-controller behavior
  • May trigger: skill-evolution-engine

Constraints

  • Must not bypass security or compliance
  • Must not infer intent without evidence
  • Adaptation must be incremental and reversible

Adapt to serve, never to shortcut safety.