surveylisted

# Survey — Multi-Perspective Codebase Analysis Run a structured survey of a codebase from multiple agent perspectives. Each perspective examines the same code at a different zoom level, produces findings as beads (tagged with `--actor`), and generates mermaid diagrams from the actual code structure. ## Arguments $ARGUMENTS Arguments format: `[repo-path] [--perspectives <list>] [--diagram-only] [--dry-run]` - **repo-path**: Path to the repo to survey (default: current working directory) - **--perspectives**: Comma-separated list of perspectives to run (default: all). Options: `prod`, `staging`, `dev`, `feature`, `pm` - **--diagram-only**: Skip bead creation, only generate mermaid diagrams - **--dry-run**: Show what would be created without creating beads If no arguments provided, survey the current working directory from all perspectives. ## The Model The survey is a **graded filter bank** — each perspective is a bandpass filter examining the same codebase at a different frequency: | Perspective | Zoom Level | What It Looks For | Actor Tag | |-------------|-----------|-------------------|-----------| | **Dev** | High freq — individual functions/files | Dead code, unused imports, TODO debt, complexity hotspots | `dev-agent` | | **Prod** | Mid-high — module/package patterns | Code smells, anti-patterns, dependency direction violations, error handling gaps | `prod-agent` | | **Staging** | Mid — test/code correspondence | Fake tests, coverage gaps, test-prod divergence, m