Analyze Phase Executor

Analyze Phase Executor

Context Tier: MINIMAL

Input

Receives via prompt:

• plugin_path: Path to plugin directory (string only)
• focus_area: context|orchestration|handoff|all

NOT PROVIDED (context isolation):

• Plugin file contents (analyzer will fetch)
• Agent file contents (analyzer will fetch)
• Session history or previous analyses
• User's other plugins or projects

Execution

Phase 1a: Initialize State

python scripts/state_manager.py init "$plugin_path" --focus "$focus_area" --mode standard

Phase 1b: File Discovery (I/O Phase)

Launch file-discovery-executor to discover files and load priorities:

Task: Discover plugin files and load priorities
Agent: coordinator-internal/phases/file-discovery-executor.md
Prompt:
  plugin_path: [path]
  patterns: ["**/*.md", "**/*.json"]

This executor:

• Discovers all plugin files (via Glob)
• Registers file references in cache
• Loads priority files (plugin.json, CLAUDE.md, entry agents)
• Returns file_refs summary

Phase 1c: Analysis (Pure Analysis Phase)

Launch analysis-executor to analyze cached data:

Task: Analyze plugin structure from cache
Agent: coordinator-internal/phases/analysis-executor.md
Prompt:
  plugin_path: [path]
  focus_area: [focus]

This executor:

• Accesses files from cache only (NO direct I/O)
• Lazy loads additional files as needed
• Launches plugin-analyzer for deep analysis
• Stores results in state

Output

analysis_complete:
  analysis_id: "[session_id from state]"
  summary:
    plugin_name: "[name]"
    violations_count: [count]
    opportunities_count: [count]
    metrics:
      agent_count: [count]
      tier_compliance: [0.0-1.0]

State Integration

• Reads: Nothing (initializes state)
• Writes:
  • immutable: plugin_path, focus_area, session_id
  • mutable.analysis_summary: Full PluginAnalysis from analyzer