---
name: assess
description: Scan generated episodes through assessment lenses and report issues with severity rankings.
allowed-tools: Read, Glob, Bash
argument-hint: "[project] --lens [lens_name]"
---

# /assess - Script Assessment

Scan generated episodes through specific "lenses" and report issues with severity rankings.

## Usage

```
/assess [project] --lens [lens_name]
/assess [project] --lens [lens_name] --batch [N]
/assess [project] --lens [lens_name] --ep [N]
```

**Examples:**
```
/assess leviathan --lens visual_richness              # Full series scan
/assess leviathan --lens visual_richness --batch 11   # Episodes 51-55
/assess leviathan --lens continuity --ep 23           # Single episode
```

## Parameters

| Parameter | Required | Default | Description |
|-----------|----------|---------|-------------|
| `project` | Yes | — | Project folder name |
| `--lens` | No | visual_richness | Which assessment lens to apply |
| `--batch [N]` | No | — | Specific batch of 5 episodes (1-12) |
| `--ep [N]` | No | — | Single episode number |

**If no batch/ep specified:** Scans entire series (60 episodes)

## Available Lenses

| Lens | What It Checks |
|------|----------------|
| `behavioral_dna` | On-screen behaviors, character specificity |
| `voice` | Dialogue distinctiveness, swap test, contamination |
| `texture` | Emotional variety, surprise, earned moments |
| `relationship_earning` | Demonstrated vs declared connections |
| `visual_richness` | Filmability, camera angles, texture, sound design |
| `continuity` | Cliffhanger chains, character state, knowledge tracking |
| `double_view_continuity` | Dialogue/action matching in double-perspective stories |

## Output

Reports issues with:
- **Severity:** MUST FIX vs COULD IMPROVE
- **Line references:** Specific locations in episodes
- **Fix suggestions:** What to change

**Deviation History Integration:**
The assess agent also checks `state/deviation_log.json` for:
- Patterns in flagged deviations across episodes
- Recurring auto-fix triggers (may indicate systemic issues)
- Episodes with multiple deviation attempts

**Note:** This agent produces reports only. Use `/rewrite` to make changes.

## Execution

When this skill is invoked:

1. **Read the agent prompt:** `/agents/assess_agent.md`
2. **Read the lens file** from `/lenses/[lens_name].md`
3. **Read target episodes** from `/[project]/scripting/episodes/`
4. **Generate assessment report** with severity rankings